William Sayers

2009
Milestone 3

05025397

Genetic Algorithms and Neural Networks

Genetic Algorithms and Neural Networks
William Sayers (05025397)
Supervised by: Colin W. Morris (BSc, MSc)
Backpropagation neural networks are usually trained using some iterative method derived from a mathematical analysis of the operation of the network. This technique is time consuming and requires understanding of the mathematical principles involved. This project will investigate the application of genetic algorithms to replace the "normal" training of the network. This involves setting up some candidate networks and then picking the best of these. The best of the population are then combined in some way to produce a new population of candidate networks. This procedure is continued until a satisfactory network is obtained.

William Keith Paul Sayers225 Page 1 of Faculty of Advanced Technology 22/04/2009

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3

Contents
Contents ............................................................................................................................ 2 1 - Introduction ............................................................................................................... 10 2 - Research .................................................................................................................... 10 2.1 - Neural Networks ................................................................................................. 10 2.1.1 - The Biological Neuron................................................................................. 10 Figure 1 - A human neuron ............................................................................................ 10 2.1.2 - The Artificial Neuron................................................................................... 11 Figure 2 - An artificial neuron ........................................................................................ 12 Equation 1 - Summation of the inputs and weights ........................................................ 12 Figure 3 - The sigmoid function shape ........................................................................... 13 2.2 - The History of Neural Networks ........................................................................ 13 2.3 - Advantages and Disadvantages of Neural Networks ......................................... 14 2.3.1 - Advantages................................................................................................... 14 2.3.2 - Disadvantages .............................................................................................. 15 2.4 - Current Applications of Neural Networks .......................................................... 17 2.4.1 - Neural Networks in Medicine ...................................................................... 17 2.4.2 - Neural Networks in Business ....................................................................... 18 2.4.3 - Object Trajectories....................................................................................... 19 2.4.4 - Robot Control .............................................................................................. 19 2.5 - Back-propagation ............................................................................................... 20 2.5.1 - Back-propagation overview ......................................................................... 20 Page 2 of 225

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3 2.5.2 - Back-propagation in depth ........................................................................... 20 2.5.3 - Back-propagation library ............................................................................. 22 2.6 - The FANN neural network library (Nissen, FANN) .......................................... 22 2.7 - Genetic Algorithms ............................................................................................ 22 2.7.1 - History of Genetic Algorithms .................................................................... 23 2.7.2 - Advantages and Disadvantages of Genetic Algorithms .............................. 24 Figure 4 - local optimum (units are undefined in this case). .......................................... 27 2.8 - Training Neural Networks with Genetic Algorithms ......................................... 28 2.8.1 - Determining Weight Values with Genetic Algorithms................................ 28 2.8.2 - Representation of a Neural Network within a Genetic Algrithm ................ 28 2.8.3 - Using Genetic Algorithms to Determine Neural Network Structure ........... 29 Figure 5 - A feed-forward neural network ..................................................................... 30 2.9 - Cascade Correlation............................................................................................ 32 2.10 - C# User Interface Programming ....................................................................... 37 2.11 - Writing C++/CLI Wrappers for Unmanaged C++ code................................... 37 2.12 - Application of Research ................................................................................... 38 3 - Design ....................................................................................................................... 38 3.1 - Program Requirements ....................................................................................... 39 3.2 - Design of the Class Structure ............................................................................. 39 3.3 - Linking C# code to managed dll’s...................................................................... 40 3.4 - Design of the User Interface ............................................................................... 40 Page 3 of 225

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3 3.4.1 - Main Form ................................................................................................... 40 Figure 9 – Network and output tab design (Main form) ................................................ 41 Figure 10 - Mean squared error tab design (Main form) ................................................ 42 Figure 11 - Dataset tab design (Main form) ................................................................... 42 3.4.2 - New Network Wizard .................................................................................. 43 Figure 12 - Design of the training algorithm tab (Wizard form) .................................... 43 Figure 13 - Design of the network settings tab (Wizard form) ...................................... 44 Figure 14 - Dataset selection tab design (Wizard form)................................................. 44 3.4.3 - Working Form ............................................................................................. 44 Figure 15 - Working form design (Working form) ........................................................ 45 3.4.4 - About Form .................................................................................................. 45 Figure 16 - About form design (About form) ................................................................. 45 Figure 17 - Dataset display design (About form) ........................................................... 46 Figure 18 - Licenses display design (About form) ......................................................... 46 4 - Implementation ......................................................................................................... 46 4.1 - User Interface Implementation ........................................................................... 47 4.1.1 - Main Form ................................................................................................... 47 Figure 19 - Network and Output tab............................................................................... 47 Figure 20 - Mean squared error graph tab ..................................................................... 48 Figure 21 - The dataset tab ............................................................................................. 49 4.1.2 - New Neural Network Wizard ...................................................................... 50 Figure 22 - Learning Algorithm tab ............................................................................... 50 Figure 23 - Network Settings tab.................................................................................... 51

Page 4 of 225

......................................................Testing Data .............1.......................................Genetic Algorithms ...................................Cascade Correlation ................ 60 5...............2.................................6 ............4 ................................... Licenses view ..................Number of iterations to achieve target (XOR) ...............................5 .........The mean squared error at testing (XOR)..........Keeping the User Interface Active whilst Processing is Occurring..................2 ........... 54 4......... 54 4...........Data Parser Implementation ........................................................Working form .................. 56 4........................... 55 4..............1 ....... 61 Figure 29 ...........1......................................................Neural Network Implementation ......Back-propagation and Cascade Correlation Implementation .....................1.......................1...4 ....... 62 Page 5 of 225 .......................................3 ........About Form ...................... 58 5........................................................................1............................................................................................................... 53 Figure 27 .......Dataset tab ..........About form.............About form.............4 .........XOR ......................................1.............................................................................. main view ...........................3 ...........................................................William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Figure 24 ...............1.....................Passing Information between Forms ................... 54 Figure 28 .................................1 ........................ 52 Figure 26 ............Graphs ..............................Working Form ........................... 54 4................................ 52 Figure 25 .............................. 57 5 .............................................About form..Back-propagation ..1....... 56 4................... Datasets view ..... 61 Figure 30 ............................. 51 4.3 .................................................................. 58 5.. 59 5....... 59 5....... 61 Figure 31 ..................................................................Genetic Algorithm Implementation .................. 57 4.....2 .......................Mean squared error upon last training test (XOR) ...........1 ..................

..Genetic Algorithms ............1... 67 Figure 37 ................................... 63 5..............................................2....................................................................................1 ... 68 6.......... 67 Figure 36 ..............................3.........2...............................................Graphs ...................Comparisons ............................................................3 .Fishers Iris data....................................................................The mean squared error at testing (Iris data) .. 64 Figure 33.....................................Back-propagation and Genetic Algorithms .........................................Virus Classification ................................................................Mean squared error upon last training test (Viruses)...1...................................................................................3 .......................................... 68 6....................................................................................................2.................... 65 5.... 69 6...........................Cascade Correlation and Genetic Algorithms .................................................Cascade Correlation .....................3.1..4 ...................2 .......... 66 5.............................1 ...Mean squared error at testing (Viruses).............................................. 69 6...................Mean squared error upon last training test (Iris data) .XOR ........ 65 5............................................. 63 5....... 67 Figure 35 ..1 ................ 63 5................ 68 6...............William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 5..................2 ................3 ...............Number of iterations to achieve target (Viruses) ..................Number of iterations to achieve target (Iris data) ............. 66 5...................4 ................... 68 6 ....................2 ..........................................................Back-propagation ....... 64 Figure 34..........3 ........Genetic Algorithms ...... 64 Figure 32 ......................................Fishers Iris Data .....3...............................Cascade Correlation ....................2 ...............................1 .............2 ...... 69 Page 6 of 225 ..........Graphs ...........................................Virus Classification.........................................................2...............Back-propagation ................3........................................ 63 5...... 66 5...........

............. 70 7............................Objectives .............1......... 73 9 .................................3 ..... 70 6...... 75 9...................................UML Class Diagram.......Evaluation .........................2 .............................................. 204 9.. 214 Page 7 of 225 ............Libraries Researched ...............4 ..................Possible User Interface Improvements .......................................................................................2 .................... 69 6.2......................... 74 9................... 69 6................. 73 8.............3 ........................................................1 ..................................................................................................1 ........Testing conclusions ... 210 10 .................................................................. 209 9.......William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 6...........................................................................................4 ..... 211 11 .....Controller ... 72 8...........Possible Improvements ........................3 .2.......1 .......How many of the original objectives were achieved? .............DLL Wrapper for FANN functions ..................Source code ................................................................................................................................................................. 174 9...Fishers Iris Data ......XOR ............ 74 9................2 ..........................................1 ...............................................................................Model .......View ......................................2 ............................... 207 9..................Appendices .....................2...... 71 8 ..................................................................5 ...........................................................................1......................................................................................1....................................................................1......Datasets............................................................Bibliography........................... 158 9........................................1 ......................................Possible Training Algorithm Improvements ................................3 ................ 70 7 ...Works Cited ................................................. 109 9....................................................Virus Classification...............................................

.......................................................................................................................... 220 13 .....................................................Tables ....................................... 217 12..................................................Table of Figures .....Table of tables .................................................................................Diary............................Table of Equations .......................................................1 ................... 222 Page 8 of 225 ............ 219 12...................................William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 12 .2 ........ 217 12..........3 .....

........ ……………………………….......... and that neither this project............................................ ………………………………............ except where specific reference is made........ Student Name (Printed) Student Signature Registered Scheme of Study Date of Signing ………………………………..... and I fully realise the consequences of plagiarising any of these sources......William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 STATEMENT OF ORIGINALITY SCHOOL OF COMPUTING DEGREE SCHEME IN COMPUTING LEVEL THREE PROJECT This is to certify that.............. the work described within this project is the result of the investigation carried out by myself.......... Page 9 of 225 ... has been submitted in candidature for any other award other than this being presently studied..... nor any part of it... ………………………………... Any material taken from published texts or computerised sources have been fully referenced..............

but in combination with a network of other neurons within the brain. the report will investigate.1 . This report encompasses design. 2. on the course “Computer Games Development”. In addition.The Biological Neuron The human brain is composed of many millions of nerve cells called “Neurons”. compare and contrast three common learning algorithms that are applicable to neural networks. Figure 1 . they can produce immensely complex results.1 . Neurons do not make decisions alone.Neural Networks Neural networks are networks of small-specialised units that sum the products of their inputs and the weights on those inputs.1. Although they are simple in concept. 2 . development and final prototyping of the program developed for this project.Research 2.A human neuron Page 10 of 225 .Introduction The purpose of this report is to bring to a conclusion my final year project on neural networks and genetic algorithms.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 1 .

or to another part of the program. a number of weights associated with those inputs. (Bourg & Seeman. Axon branches connect to synapses to form the neural networks within the brain. as mentioned above). This action potential extends down the axon to the cells at the end of the axon branches (often synapses connecting to other neuronal cells. in order to allow the brain to control cells external to the brain.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 A neuron is composed of a cell body.1. the nucleus of the neuron cell processes them and produces an output. If the inputs from the dendrites to the neuron nucleus when summed are past a certain threshold. Page 11 of 225 . In the case of the axon branches connecting to synapses that are part of a second neuron. Concerning inputs from the dendrites.The Artificial Neuron In order to construct an artificial neuron we need a way of representing the same model in data structures and algorithms. an activation function and a method for representing an output – either to another artificial neuron in the next layer. a synaptic gap occurs between the axon branch and the synapse. 2006) 2. 2004).2 . The axon and dendrites both split before terminating into “Axon branches” and “Synapses”. (Barber. Axon branches can connect to other types of cells also. AI: Neural Network for Beginners pt. 1. the nucleus will generate an “action potential”. The usual method for approaching this is to have a number of inputs. with extensions called “Dendrites” which are the neuronal inputs and a single extension called an “Axon” which is the output for that neuron. a summing function. in this gap between neurons chemical reactions take place that will inhibit or excite the signal.

Page 12 of 225 . The product of the data item and its associated weighting is then stored before the summation of all the data items and associated weights occurs.An artificial neuron The system devised with these components is as follows. or integer). or one (activated) if the summed products are above a certain value.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Figure 2 . ℎ = × ℎ Equation 1 . The activation function accepts this stored value. A specific input accepts a data item (usually a binary digit. real number. A stepping function can be used which will output zero (or non-activated) if the summed products are below a certain value. a differentiable function of some kind (such as a sigmoid function) which will output values between zero and one depending on the value of the inputs is another possibility.Summation of the inputs and weights The next stage depends on the activation function chosen. Alternatively.

(Bourg & Seeman. 2004).The History of Neural Networks The foundation of neural network research are in psychology. 2004). McCulloch and Pitts formulated the first neural network model which featured digital neurons but the neurons had no capability to learn (McCulloch & Pitts. In this example “y” is the function output and “t” is the function input.4 0. as defined in the graph above. 1. however he was unable to come up with a reliable mathematically accurate mechanism for allowing multilayer perceptrons to learn (Rosenblatt. in 1958.2 1 0.The sigmoid function shape = 1 1+ Equation 2 . Frank Rosenblatt developed the perceptron model then.8 0. AI: Neural Network for Beginners pt. 1958). (Barber. contributed to the ideas that underpinned early neural network research. 1943). 2006) 2.2 0 -6 -4 -2 0 2 4 6 Figure 3 . Page 13 of 225 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 1.A sigmoid function (Bourg & Seeman.2 . the work of 19th century psychologists such as Freud. William James and others.6 0.

1974) discovered the back-propagation algorithm (independently rediscovered by Parker in 1982 (Parker. 2.Advantages Neural networks make possible or practical many things that with conventional solutions would be extremely difficult.Advantages and Disadvantages of Neural Networks 2.3. 1992). of “n” nodes.1 . The individual neurons in a neural network have no time dependencies on each other and each can therefore be run on a separate processor (or as separate threads on a single processor) if desired without causing the problems often associated with such parallelism. Back-propagation allows for the training of multilayer perceptrons and in particular. A good example of this is that neural networks have been implemented to run on architecture originally designed for processing three dimensional computer graphics. used on this architecture they become an extremely fast as well as flexible tool. Werbos discovered the algorithm whilst working on his doctoral thesis in statistics and called it “dynamic feedback”. (Blum. when Werbos (Werbos. taking Page 14 of 225 . where “n” is an undefined number. perceptrons with “n” hidden layers. One of the benefits of neural networks when compared with other problem solving techniques in that they are inherently parallel and thus can run very effectively and efficiently on parallel hardware. The next major advancement in neural networks was not until 1974. 1982)).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Two layer perceptrons were the subject of several experiments to determine their usage and effectiveness. Parker discovered it independently in 1982 whilst performing graduate work at Stanford University in the United States of America and called it “learning logic”.3 .

2001).The statistical approach to deciding relevant data Another advantage of neural networks is that noisy data is not a real problem for them to learn to interpret.The statistical approach to deciding relevant data”. in fact. to attempt to provide a more robust post-training network. where explaining why you have arrived at a particular diagnosis is an important part of the process. 1 Decide on relevant data.Disadvantages One disadvantage to neural networks can be that it is very difficult for a human being to analyse a trained neural network and describe why it may come up with one answer over another. 2. 2 Formulate a statistical model. We can simply present the neural network with all the data as opposed to following the circuitous and overly complex (by comparison) statistical approach demonstrated in “Table 1 .3. noisy data is utilised during training upon occasion.2 . 5 Build a system that incorporates what we have learned. 4 Analyze the results.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 advantage of the massively parallel architecture of this hardware and the programmable sections of the graphics pipeline (vector and fragment shaders) to produce extremely fast neural networks (Davis. Page 15 of 225 . Table 1 . as irrelevant data will simply end up weighted so that it has zero or close to zero actual effect on the solution produced by the network. 3 Run the formulated model. This is a large disadvantage in areas such as medical diagnosis. When using a neural network there's no need to establish before attempting problem solving which data is relevant. Neural networks can excel at determining what data in a particular set is relevant.

The training times on neural networks can also be a disadvantage. a sunny day. They collected one hundred images with a concealed tank. then a further one hundred images without. or even to verify the quality of the training. Therefore. Fifty pictures of each set were set aside and the neural network trained with the remaining one hundred images with and without tanks. it can also be hard to tell how well (or how badly) trained a neural network is. All the photographs with tanks were from a cloudy day and all the images without. requires largely heuristic construction and preparation. The US military then tested the neural network with a further set of one hundred images. the network had trained itself to recognise whether the sky was cloudy or not. train the neural network. there is no formal methodology with which to choose the architecture. Page 16 of 225 . asked to differentiate between the two sets of pictures. After the neural network’s training was complete. A classic example is that in the 1980s. the US military wanted to use a neural network to analyse images and determine whether they contained a concealed tank. When using neural networks is that. A tool used to solve heuristic problems. for obvious purposes. with and without tanks.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Once a neural network’s training is complete it is hard to tell why it is coming up with a particular solution. although the length of time the problem would take to solve without employing a neural network must be taken into account when deciding how much of a disadvantage. Eventually the original data set was re-examined. the network came up with apparently random answers. they tested the network with the remaining fifty images that the neural network had not seen before and the neural network correctly identified most of the tanks. only to find that on this round of testing.

the usual best approach is to simply feed them as much data as can be obtained. the more accurate and in line with expectations and desires. Fortunately since neural networks are (as has been mentioned above) good at ignoring irrelevant data. Testing may then take place with the “testing” set to determine accuracy with unseen data. 2. “training” and “testing”. the eventual result will be. The real strength of neural networks. One of the main research areas is using neural networks to recognise diseases/conditions via scans. Garbage out”. covering as many relevant situations as possible.Current Applications of Neural Networks 2. this has to be done within reason however. This is generalisation and it is far more useful than it at first appears and similar from a very high-level perspective to how the human brain and brains in general function.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Another possible criticism is that neural networks are extremely dependant on the quality and content of the original training data set. Data sets for training neural networks generally separate into two sub-sets. The more time spent training the neural network on quality data sets.Neural Networks in Medicine There is a large amount of research now covering the applications of neural networks in medicine.4. or it does exacerbate the problem with the time spent training the neural network. lies in their ability for generalisation. Neural networks make it more apparent however because (as in the previous US Military example) it can be difficult sometimes to decide what constitutes good data.4 . This drawback is however one that is generally universal to computing (and many other areas) often referred to as “GIGO” or “Garbage in. The network training first takes place using the appropriate “training” set. Page 17 of 225 . in the opinion of the author.1 .

and assessing borrowers have all incorporated neural networks as an integral part of the system. A back-propagation neural net is integrated with the airline marketing tactician which monitored and recommended on booking for each flight. Businesses have used neural networks in the past for applications such as assisting the marketing control of airline seat allocations in the AMT (Airline Marketing Tactician). thus supplying information more or less directly linked to the airlines main income. However. Credit scoring. being good at analysing patterns and predicting future trends. (Stergiou & Dimitrios. 1997) 2. research on using neural networks to diagnose diseases took place. You can also use the technology to model parts of the human biological system in order to better understand it.Neural Networks in Business Neural networks. merely a variety of scans covering all possible permutations of the disease or condition. for example the human cardio-vascular system. In the 1980s. (Stergiou & Dimitrios. mortgage screening. can fit very well into most business situations. The main reasons for using a neural network to model the human cardio-vascular system are that a neural network is capable of adjusting itself to be relevant to a particular individual and also to adapt by itself to changes in that individual.4.2 . a clear problem with this approach is that when using a neural network it would be difficult to tell how a neural network reached its conclusion.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 This is particularly suitable work for neural networks because you do not need perfect examples of how to recognise the diseases and conditions. 1997) Page 18 of 225 .

William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 2.Robot Control Controlling manipulators (such as robotic arms in a car manufacturing plant) is usually how robotics and neural networks connect. A neural network allows the robotic arms to be very flexible in their operation and perform self-adjustments as time goes by. 1999) 2.4. velocity and acceleration of the object being estimated in those implementations by several neural networks using several of the most recent measurements of the object coordinates.3 . The position. When controlling manipulators a neural net must deal with four major problems (Robot Control. (relatively speaking) involving the following steps (Robot Control. 2008): • • • Determine the target coordinates relative to the robot. When a robot arm is one hundred percent accurate that is a simple task. compensating automatically for wear and tear on Page 19 of 225 . (Marshall & Srikanth.e.Object Trajectories Predicting the trajectory of an object in real time is another area where neural networks have been utilised.4 . a robotic hand or any other similar manipulator) in the appropriate position to grasp or otherwise manipulate an object or part of an object. Adjust the arm to the appropriate joint angles and close the effecter. Calculate the joint angles required for the arm to be in the appropriate position.4. 2008): • • • • Forward Kinematics Inverse Kinematics Dynamics Trajectory generation The ultimate goal of course is to position the “effecter” (I.

A commonly used function is the sigmoid function as described in section 2. This rule essentially describes adjusting the weight using the difference between the expected and actual outputs. The advantage of using a function like this is that it allows us to differentiate how close we came to the correct result. Assign responsibility to each of the neurons in the previous level. In order to use a back-propagation training algorithm you must have a non-linear activation function for your artificial neuron.1. Adjust the weight of the output neurons to lessen the error. The sigmoid function is in the source code at 9.Back-propagation overview In essence.2 .5.1 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 themselves.Back-propagation 2.3. 2.5 .5.2 -The Artificial Neuron. 2008). 2.1 -“Network. allowing us to apply our learning algorithm. Adjust the weight of the neurons to minimise the responsibility.Back-propagation in depth In a single layer neural network (or with a single node) the perceptron rule is applicable. Page 20 of 225 . training by back-propagation involves the steps: • • • • • Present a set of training data Compare the network’s output to the desired output at each output neuron and calculate the error.cs” lines 130-133. This is far preferable to having to perform lengthy and expensive re-calibrations on any regular basis (Robot Control. the disadvantage of a differentiable function is that it is by the use of these functions that local optima become apparent in the search space. based on the strength of the weights connecting them to the output neurons.1.

Back-propagation for the Uninitiated.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 ∆ = ℎ = − . Back-propagation for the Uninitiated. ℎ : = [1 − ] . The sigmoid function differentiates very neatly. the output layer nodes training function can be written as: = 1− − Equation 6 . “n” is the learning rate of the neural network.Altered delta calculation rule for the output layer (Matthews. the following alteration to the perceptron learning rule would serve our purposes (Matthews. 2002) Calculating the hidden layer deltas is a little more complicated.1. Equation 3 – The delta rule (Matthews. The effect that the change will have on the following neurons in the network is an extra factor. Back-propagation for the Uninitiated. The other two values “di and yi” are the actual and desired outputs.differentiated sigmoid function (Matthews. this rule has no positive effect in a multi-layer neural network. such as the sigmoid function (see 2. as the effect that the change in the weight will have on the rest of the network is a missing factor. Back-propagation for the Uninitiated. respectively. 2002) However.2 -The Artificial Neuron) an alteration can be made to the perceptron learning rule allowing this difficulty to be overcome. 2002) In this equation “w” and “x” are as before. 2002): ∆ = ℎ = − Equation 5 . Provided you are using a differentiable function. Assuming the sigmoid activation function is in use. Page 21 of 225 . 2002) Therefore. = 1− Equation 4 . Back-propagation for the Uninitiated.Altered perceptron learning rule (Matthews.

William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Equation 7 .6 . 2002) As demonstrated above. 2. FANN) The FANN library (see (9. in a language that can be linked to C# fairly easily and it also supports cascade neural networks. although as it turned out I did not). Back-propagation for the Uninitiated.Altered delta calculation rule for the hidden layer(s) (Matthews. in order to apply this learning mechanism you start at the output of the network and work your way towards the inputs. it is fast. By using similar techniques to natural selection. 1859) very neatly avoids one of the larger problems involved in software design: how to specify in advance all the possible permutations of the original problem and how the program should react to those permutations. it is possible to “breed” solutions to problems. to calculate the delta for a hidden layer requires the deltas for the following layers.3 .Back-propagation library 2. open-source (allowing me to alter it if I needed too. easy to implement. The FANN library was decided upon for several reasons. saving me finding a separate library.7 . Therefore. 2.3 -Libraries Researched) for other libraries investigated) implements a back-propagation solution and this library will be used in the project to implement a feed-forward back-propagation neural network.The FANN neural network library (Nissen. Most organisms evolve by means of sexual reproduction and natural selection (Darwin. Natural selection (Darwin.5. against which to compare the genetic algorithm trained neural network. 1859) Page 22 of 225 .Genetic Algorithms Genetic algorithms attempt to copy natural laws to a certain extent in order to apply a random search pattern to a defined search space.

2. then of the technical University of Berlin introduced a technique he named evolution strategy. this mixing leads to much faster evolution than if off-spring simply copied the genetic material as it is in the parents. however. occasionally causing the genetic information. simulating processes of biological evolution. By the early 1980s. This could result in either an effective or defective solution. swapping genetic material. including noisy and discontinuous data. built on papers both by himself and by other researchers. Also in 1975. across a range of subjects.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Sexual reproduction ensures mixing and recombination of genes. a dissertation by Kenneth De Jong established the potential of genetic algorithms by showing that they perform well on a wide variety of data. In 1965 one Ingo Rechenberg. At first. This book presents the concepts of adaptive digital systems using mutation. during this process. the publication of the book “Adaptation in Natural and Artificial Systems” occurred.1 . During the process chromosomes line up together and then cross over partway along their length. Mutation plays a part in this as well. genetic Page 23 of 225 . many researchers had independently developed evolution-inspired algorithms for function optimisation and machine learning but did not get much follow up to their work. by the author John Holland. selection and crossover. this application was largely theoretical.History of Genetic Algorithms Genetic algorithms first appeared on computers in the late 1950’s and early 1960’s. genetic algorithms were becoming more widely used. but this variety helps to prevent stagnation in the gene pool. mostly designed to help evolutionary biologists’ model aspects of natural evolution. By 1962.7. although this technique involved no crossover. or indeed multiple genomes in a population. to alter slightly in a random way. In 1975.

proceeding with other more likely candidates. Genetic algorithms effectively explore many different branches of the tree at once and when a certain branch turns out to be non-optimal.1 .2 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 algorithms soon moved into commercial territory and nowadays help solve many problems in many different areas. This leads to the exploration of a large proportion of the search space.2. These are “non-linear” Page 24 of 225 . abandon that search.7.Advantages and Disadvantages of Genetic Algorithms 2. Thanks to the parallelism that is a genetic algorithms main advantage. (Marczyk. This is “Schema Theorem” and allows a genetic algorithm to move towards the searchspace with the most promising individuals in a timely fashion and then select the best member of that group (Marczyk. or further up the hierarchy. more conventional. methods. they are very well suited as a means of exploring search spaces too large to be effectively searched in a reasonable amount of time by other.Advantages The primary advantage of genetic algorithms is their inherent parallelism.7. in the same way that the average response of a relatively small percentage of the population of any given country. Many other algorithms are largely serial and can only explore a search-tree so far in any one direction before abandoning all progress and starting again from the beginning. with processing power exponentially dedicated to better areas as the exploration progresses. 2004) 2. The real advantage of parallelism however is that by evaluating the relatively small search-space that it does. a genetic algorithm is implicitly evaluating a much larger group of individuals. can be used to fairly accurately predict national trends. 2004).

In a linear problem. such as crossover and mutation. Genetic algorithms have no prior knowledge about the problem they are trying to solve. One of the largest strengths of genetic algorithms can at first glance appear to be their largest weakness. A good example of this is the concept Page 25 of 225 . each searching their own space for the best solution. each component’s fitness is individual and any improvement to individual component's fitness is necessarily an improvement to the whole. This allows them to discover solutions that other algorithms may have over-looked. This non-linearity results in a huge increase in the search space. due to their strength in navigating large search spaces (Marczyk. it is hard to know whether we have reached the global optimum. they tend towards non-linearity where altering one component's fitness positively may make other components less fit and cause a ripple effect across the whole system. the random nature of the various starting points of the initial population and the other methods they employ. 2004). they produce random changes to their candidates and then use the objective function to determine whether those changes are positive or negative overall. crossover allows promising solutions to share their success and mutation allows random changes in the local search space of a given solution. making genetic algorithms an effective way to search them. genetic algorithms tend to give good results compared to other search strategies. Many real life problems are not like this. A third advantage of genetic algorithms is that they do not tend to be easily trapped by local optima. or never contained in their search space in the first place. selection and mutation a genetic algorithm is metaphorically similar to a large number of parallel depth first algorithms. However.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 problems. or merely very good local optima. In infinite or very large search spaces. Without crossover. due again to the parallelism of their approach. Selection allows the pruning of the least promising searches.

2004). 2004). Generally. therefore no gradual increase can lead you to it. Page 26 of 225 . Likewise. a comparatively smaller segment of the search space will be covered and the eventual solution may take longer to reach or never be reached as a result. if the size of the population is too low.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 of negative feedback. the genetic algorithm may end up solving an entirely different problem from the originally intended one. with each number representing some distinct part of the solution (Marczyk. the system will never converge towards a suitable solution. For example in the diagram below. however that is only one point. the global optima is at the far right. where the actual code of the program forms the genomes in the genetic algorithm and it can swap portions in or out. A third issue is setting the correct mutation rate. there is no slope leading to it. rediscovered by genetic algorithms.2 . integer. A further problem is that fitness functions can be deceptive. then genetic algorithms (and most other search techniques) are no better than a random search for finding the solution. the investigation may cover too little of the search space to find the optimum solution (Marczyk. If you have a poorly written fitness function. population size. 2004).Disadvantages One of the disadvantages of a genetic algorithm is the necessity of representing the problem the genetic algorithm is trying to solve. etc. genetic algorithms use strings of binary. Another method is to use genetic programming. however if it is too low. or real-valued numbers. or adjust them.7. 2. in a form that the genetic algorithm can use. A second disadvantage of genetic algorithms is that the fitness function is crucial to the development of a suitable solution for the problem. if the solutions to various problems are in areas of the search space that the fitness function would not consider likely. but denied a patent for several years because it ran counter to established beliefs (Marczyk.2. If the mutation is too high.

Page 27 of 225 . A further problem is premature convergence: this occurs when a mutation early in the process produces a large leap in the fitness of that particular genome. local optima in the centre of the diagram. This is because analytical solutions usually produce more accurate results faster than a genetic algorithm (or any heuristic method) is capable of (Marczyk. In this example a properly constructed genetic algorithm would be likely to settle on the marginally less optimal. 2004). which will then reproduce abundantly. where analytical solutions exist they should be take precedence over genetic algorithms. Finally. 900 800 700 600 500 400 300 200 100 0 0 2 4 6 8 10 12 Figure 4 .local optimum (units are undefined in this case). including sigma scaling and Boltzmann selection (Mitchell. unless by random chance it discovered the point with the most optimum solution. There are various methods for solving this. but altogether easier to find. 1996).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 you either have the solution or you don’t. lowering the diversity of the population and resulting in genetic algorithms possibly falling into the local optima that the mutation represents.

a neural network representation within a genetic algorithm is a concatenation of all the weighs in the neural network.7 -Genetic Algorithms) are good at avoiding being stuck on local optima due to their searching several regions of search space simultaneously.8 . however.Determining Weight Values with Genetic Algorithms Genetic algorithms (as described in 2. these are not a complete solution (Blum.8. involving finding the optimum values for a set of real numbers (connection weights) which will produce the least error in the results from the neural network. In addition.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 2. the only information genetic algorithms require to train a neural network is an objective function (as described in 2. Genetic algorithms also have the benefit that they do not place any restrictions whatsoever on the architecture of the neural network. 1992). such as adding a “momentum” to the algorithm. since you are merely increasing the size of the genomes being worked with.Representation of a Neural Network within a Genetic Algrithm As a rule. to try to help it move beyond local optima.Training Neural Networks with Genetic Algorithms Training neural networks is a form of problem solving. not altering them besides that in any way. Page 28 of 225 . however the method of choice for training neural networks is usually back propagation (see 2.5 -Back-propagation) which is a gradient descent algorithm and as such can be easily trapped in local optima (Branke.8. There are methods to try to avoid this problem when using back propagation.1 . 2. 1995). The error surfaces associated with these problems tend to be highly variable and contain many local optima.2 . 2.7 -Genetic Algorithms). to test the appropriateness of a given genome.

3 . outputs to inputs. can lead to very large strings (thousands of bits) (Branke. side by side). 2. it is also possible to use a genetic algorithm to evolve a structure for a neural network. The average neural network has an input layer. one or two hidden layers of an indeterminate number of nodes and one output layer. The problem at hand dictates the input and output layer structures. These are all connected.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Crossover can disrupt genes far apart. place a neurons input and output weights. Page 29 of 225 . a nodes input weights and the nodes of a layer. One of the most important decisions when deciding how to represent a neural network within a genome is whether to use binary strings or real number strings.Using Genetic Algorithms to Determine Neural Network Structure As well as using a genetic algorithm to evolve the weight set for a fixed-structure neural network.e. 1995). but is less likely to do so for genes close together. Standard genetic algorithms use binary strings. however representing a large number of real valued numbers as a binary string.8. Because of this it is helpful to keep functional units close together (I. to form a feed forward network.

However.A feed-forward neural network This architecture is then altered (number of nodes in each hidden layer. if the neural network architecture is too complex. the neural network will learn fast but will generate too specific a solution and will not generalize between similar inputs. a neural network topology to deal with that problem.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Figure 5 . given a specific problem. the neural network will never learn to solve the problem (or come close enough to suggest a solution). there is currently no method by which to establish. There is also no method to check how optimal your solution is (Branke. If neural network architecture is too simplistic for the problem at hand. a neural networks generalisation is its main strength and therefore you do not wish to lose it. As mentioned in section 1.0. 1995). this is not necessarily the most efficient form for a network for any given problem and the trial and error method is an inefficient method for determining the correct architecture for an optimally performing neural network for the problem at hand. Page 30 of 225 . On the other hand. Although network architecture plays such a large role in network performance. number of hidden layers) by trial and error and intuition to establish a neural network that performs well for the problem at hand.

or high-level encoding (Branke. 1995). a representation is required which can accommodate all networks that will work for the problem but none that will not. or low-level encoding and the weak. the genetic algorithm represents the neural network by means of its connections. In cases where a genetic algorithm determines the number of hidden layers and the number of nodes per hidden layer and the network itself is interconnected fully. you must remove all connections to or from that node completely. being able to contain the most optimum solution for the problem at hand.1 .Representing neural network structure Representing the structure of a neural network within a genetic algorithm is not as straightforward as representing the weight values of a neural network within a genetic algorithm. thus meaning that in order to remove a node completely. Ideally.8. Page 31 of 225 . Low-level encodings specify each connection individually. 1995).3. There are two methods for representing neural network topologies within a genetic algorithm.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 2. whereas high-level encodings group together connections and/or nodes. an exception is possible. In direct encoding. indirect. thus eliminating the searching of meaningless space and by definition. the strong. the network is divided into areas and for each area the number of nodes and the density of connections to other nodes is determined (Branke. direct. Back propagation learning rate is also a parameter for one of these areas. In one version of high level encoding of a neural network.

Neural networks that are too small on the other hand. 1991). These mapping methods are therefore more suitable for small networks or networks with a small number of connections. These are the Step-size problem. This section largely drawn from information in: “Curved trajectory prediction using a self-organizing neural network” (Branke. will not ever succeed in learning the problem. As mentioned in section 5. A problem with high level encoding is that regular networks are favoured but not every network is equally probable. This can lead to human bias entering the system and affecting the outcome.Cascade Correlation The cascade learning architecture is a method by which the training algorithm builds the neural network as it proceeds. This could be useful.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 One of the reasons why alternatives to low-level encoding arose is the exponential rise in the number of connections. Therefore. by limiting the potential size of the networks evolved and preventing the generation of overly large and complex networks but on the other hand.2 neural networks that are too large will solve a problem fast but will not generalize well and therefore their usefulness is limited severely. 2. as the size of the neural network increases. it is impossible to know whether these large and complex networks may be the most suitable way to solve the problem at hand. 1995).9 . which is obviously undesirable. and the moving target problem (Fahlman & Lebiere. a balance between speed of learning and generalisation exhibited in the trained network is necessary. Page 32 of 225 . The Cascade-Correlation Learning Architecture. The algorithm attempts to solve the issues associated with back-propagation that produce slow learning.

Equation 8 . After a number of training cycles pass with no substantial improvement (the precise number being determined by a user-supplied “patience” parameter). The connection between each input and output is via an adjustable weight.2 -Back-propagation in depth) or the quick prop rule (Fahlman.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 The system begins with a basic neural network of inputs and outputs (numbers of each dictated by the task as usual) and no hidden layer. the partial derivative of "s" with respect to each of the candidate unit’s incoming weights "wi" is a necessity. = −` −` . the residual output error at “o” (Fahlman & Lebiere. with the aim of getting as close to a solution as possible. The Cascade-Correlation Learning Architecture.The definition of 's' given that 'o' is the network output at which the error is measured. If a trigger occurs for the addition of a unit. 1988).5. 'v' is the candidate unit's value. The Cascade-Correlation Learning Architecture. the initiation of a number of “candidate” units takes place and trained using either the perceptron rule (2. It then trains the adjustable weights using the perceptron rule (as described in 2. In order to maximise "s". 1991). either an extra node is added. The output units may use either a linear or differentiable activation function. There is also a bias input. The “candidate” nodes take the neural net inputs. the candidate unit’s value. 'p' is the training pattern. or sometimes the quick prop rule (Fahlman. An Empirical Study of Learning Speed in Back-Propagation Networks. and attempt to maximise “s” the sum overall output units “o” of the magnitude of the correlation between “v”.2 -Back-propagation in depth). 1991). plus the outputs of any previously created units. and “Eo”. 1988). or if the networks output is of sufficient quality the training is ended.5. An Empirical Study of Learning Speed in BackPropagation Networks. the quantities ‘`v ’ and ‘`Eo’ are the values of v and Eo averaged over all patterns (Fahlman & Lebiere. Page 33 of 225 . with a value of one.

The training iterations for these adjustable weights in the main network then continue. . and the inputs for that node are frozen. Adjustable weights link the outputs for this node and all the output nodes inputs.The partial derivative of ’s’ (Fahlman & Lebiere. a gradient descent to maximize "s" can take place.The partial derivative of ‘s’ with respect to each of the candidates incoming weights ‘wi’. where ‘oo’ is the sign of the correlation between the candidates value output ‘o’. The connection of the best of the candidates to the neural network as it stands takes place. = . After these computations are complete.p’ is the input the candidate receives from unit ‘i’ for pattern ‘p’. Equation 10 . Once the correlation on a candidate has reached an acceptable level. −` ′ .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Equation 9 . Page 34 of 225 . Expansion and differentiation of this calculation can then take place. ‘f'p’ is the derivative for pattern 'p' of the candidate unit's activation function with respect to the sum of it’s inputs and ‘Ii. 1991). or the completed training epochs reaches a maximum level. The Cascade-Correlation Learning Architecture.

A first node has been added. 1991) Cascade-Correlation 1991). The Cascade Correlation Learning Architecture. 1991) Cascade-Correlation 1991).The second state of a cascade correlation neural network. Figure 7 . Page 35 of 225 . The square econd connections are locked weights (Fahlman & Lebiere. The Cascade Correlation Learning Architecture.William Sayers 05025397 Genetic Algorithms and Neural Networks Neural Milestone 3 Figure 6 . The circular connections are adjustable weights (Fahlman & Lebiere.The initial state of a cascade correlation neural network.

Page 36 of 225 . with two nodes added (Fahlman & Lebiere. The Cascade-Correlation Learning Architecture. 1991).The third state of a cascade correlation neural network.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Figure 8 .

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3

2.10 - C# User Interface Programming
C# uses the Winforms API to create user interfaces. It allows you to “draw” objects on a form in a visual manner to design your layout and then access and alter properties and subproperties of those objects in your programming code. A separate thread for each Window in Winforms exists, in order to facilitate performance on modern multi-core machines, although it does make it more difficult to move data between the forms. A method using “delegate functions” (similar to “function pointers” in C++) and “events” (Lysle, 2007) is the method I will use. Another challenge is making the neural network training independent of the user interface, in order to avoid the interface freezing whilst training a neural network. The solution I plan for this is to use the background worker object provided with C#, which allows you to assign a delegate function to be your background task, and will trigger events upon completion, for which event handlers will be executed (Microsoft, BackgroundWorker Class, 2009). Making a C# user interface interact with C++/CLI dll’s simply involves adding a reference to the dll in your project, via the Visual Studio® user interface.

2.11 - Writing C++/CLI Wrappers for Unmanaged C++ code
The cleanest method of connecting a native C++ or C dll or piece of code to a managed C# interface is via a C++/CLI bridge class. The VC++ compiled can swap between managed and unmanaged code on the fly whilst compiling and allows you to have both managed and unmanaged code (in C++) as part of the same project. Thus, a managed class can communicate with the unmanaged class (performing appropriate marshalling of the non-primitive data types (Microsoft, Using C++ Interop (Implicit PInvoke), 2009)) and the C# code can then access it directly.

Page 37 of 225

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3 Although it is possible for a C# program to interact with an un-managed dll directly, this method is less error-prone and more likely to be robust and stable, as well as making for cleaner code.

2.12 - Application of Research
The research presented here will help to produce the deliverables for milestone two and milestone three of my final year project: I plan to combine C# for the user interface and C++ for the neural network back-ends in order to create an efficient, powerful and fast solution. Section 2.11 -Writing C++/CLI Wrappers for Unmanaged C++ code, covers the combination of C# and C++ so I will not bother to cover it here, I plan to use the DLL method, with one DLL combining managed and unmanaged C++ linked into my C #user interface code. The user interface will then have methods to adjust parameters such as the genetic algorithms population size, number of generations, etc. In order to test and see what kind of genetic algorithm learns and develops fastest. It will also allow you to specify the number of back-propagation iterations. The use of this application will be in evaluating the difference in performance between a genetic algorithm, back-propagation algorithm and cascade correlation algorithm as neural network training algorithms.

3 - Design
The purpose of this application will be to allow the user to run a back-propagation trained neural network, or a genetic algorithm trained neural network and present the results of the network in such a fashion as to allow analysis and comparison of the training methods and their suitability in different situations.

Page 38 of 225

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3

3.1 - Program Requirements
The program must be able to: • • • • • • Run a genetic algorithm on a feed forward fully interconnected neural network to train the network. Run a back propagation algorithm on a similar network to train the network. Run a cascade algorithm on a similar network to train the network. Allow the user to set appropriate variables (see 3.4.2 -New Network Wizard) to adjust the execution of the learning algorithm selected for the neural network. Allow the user to select which learning algorithm they wish to use. Display the results of running that algorithm on a neural network in a meaningful fashion.

3.2 - Design of the Class Structure
Before I discovered a simple to program and effective design, I tried and discarded two designs. I eventually used the third design and it is that third design documented here. The two prior designs to this involved representing neural networks via objects and compositions of objects in vectors, which lead to flexible, but slow and overly complex code upon experimentation. The final solution, the solution documented here, is a far simpler one (in terms of the neural network representation, it is far more detailed in the other areas than the initial designs were). This solution loosely follows a model, view, controller architecture. The solution enables easy replacement of sections of code as long as the interfaces and outputs remain the same. For more information on the class structure, see section 9.2 -UML Class Diagram.

Page 39 of 225

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3

3.3 - Linking C# code to managed dll’s
Linking C# code and managed dll’s is a simple process (in Visual Studio) as explained in (2.11 -Writing C++/CLI Wrappers for Unmanaged C++ code) of selecting the references section in the C# project in question and adding the dll you wish to refer too. Then you can simply add a “using” directive to use the dll namespace and create, destroy, and use the classes within as if they were native C# classes. This can be seen in (9.1.2.6 “Algorithms\BackProp.cs”) and (9.1.2.4 -“Algorithms\CascadeCorrelation.cs”) with the “using FANN_Wrapper;” directive. This is superior to the direct C# to unmanaged code linking that used in milestone two, since it makes for cleaner and thus more easily maintainable code and the code is more robust in this fashion. The robustness comes from the fact that C++/CLI is far better at interacting with unmanaged C++ code than C# is. Indeed – they can even be in the same source file and compiled into one object. Therefore, by using a C++/CLI wrapper to interface to the unmanaged C++ and then interacting with the C++/CLI wrapper/s errors are less likely to arise.

3.4 - Design of the User Interface
The construction of the user interface utilised the Winforms based tools in visual studio 2008 (see 2.10 -C# User Interface Programming). I also used the ZedGraph control (ZedGraph) to display the graph of my results.

3.4.1 - Main Form
The main form is the crux of the application (as is default in C# Winforms programming) and if the main form is closed, all other forms and classes are disposed and the application terminates.

Page 40 of 225

William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 3.1.Network and Output tab This tab needs a method by which to initialise the adding of a new network (3. along with two displays.1. One display will show network details and one display will show the training and testing output. Figure 9 – Network and output tab design (Main form) 3.4.1 .2 -New Network Wizard) a method to train the current network and a method to test that training. which simply needs to display the graph of the mean squared error over the current epoch/generation.2 .Mean Squared Error Graph tab The mean squared error graph tab is a simple tab. Three buttons will therefore be present.4.4. Page 41 of 225 .

Dataset tab design (Main form) Page 42 of 225 .Dataset tab This tab needs two display sections. Figure 11 .4.3 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Figure 10 . to show the training data and to show the testing data.Mean squared error tab design (Main form) 3.1.

2. 3.9 -Cascade Correlation).1 -Learning Algorithm tab (if no algorithm has yet been selected.2.2 .4.4.2. the default is genetic algorithm).Design of the training algorithm tab (Wizard form) 3.Network Settings tab This tab allows you to set appropriate settings for the network and training algorithm you have selected in 3.5 -Back-propagation.1 .2 . out of three choices previously mentioned (2. The user can progress through the steps via the “previous” and “next” buttons or via the tabs at the top in whatever order you choose.New Network Wizard The new network wizard is the form that appears when the user selects the creation of a new network.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 3.2. Verification of all data takes place before the form passes control back to the main form and triggers a main form event with a parameter that takes a specially constructed class to transfer the data from the wizard. 2.7 Genetic Algorithms. Page 43 of 225 . It allows you to move in a logical progression through the set up process of a new neural network and associated training algorithm.4. Figure 12 .Learning Algorithm tab On the learning algorithm tab you select the learning algorithm for the neural network you are creating.4.

Page 44 of 225 . both to protect the main form from data inputs that may cause training problems and to provide a visual indicator of work taking place.Dataset selection tab design (Wizard form) 3.Design of the network settings tab (Wizard form) 3.Working Form The working form is merely a small display shown as a dialog (lock out access to displaying form) which shows whilst a neural network is training. or to choose custom data sets from your hard disk for training and testing data (appropriately formatted in the FANN style).4.3 . Figure 14 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Figure 13 .3 .Dataset tab The dataset tab allows you either to select from two pre-setup datasets.2.4.

Page 45 of 225 .Datasets display Clicking the data sets button displays a tabbed interface.About form design (About form) 3.4.4.Working form design (Working form) 3. project description).1 .4 . one tab with XOR information and one tab with Fishers Iris Data information. project supervisor.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Figure 15 .About Form The about form displays information about the application (author. Clicking the datasets button or the licenses button displays the appropriate information. clicking the same button again hides that information. displays licensing information and displays information on the pre selected datasets.4. Figure 16 .

4. Figure 18 .Implementation The implementation is largely in C#. FANN) library written in C and linked as a static library to the managed FANN_Wrapper dynamic link library. it will display the appropriate license.Licenses display design (About form) 4 . depending on your selection.4. Page 46 of 225 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Figure 17 .Licenses display Clicking the licenses button displays a small information panel with selection methods.2 .Dataset display design (About form) 3. with small portions in C++/CLI and the back-end FANN (Nissen.

1 .1 .cs”.Network and Output tab Figure 19 .1.2 -“FrmMain.Main Form The main forms code is in section 9. 4.Network and Output tab Page 47 of 225 .1. 4.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 4.1.User Interface Implementation The implementation of the user interface is mainly in C#. mostly via the visual design tools found in visual studio 2008.1 .1.1.

1.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 4.1.Mean Squared Error Graph tab Figure 20 .Mean squared error graph tab Page 48 of 225 .2 .

The dataset tab Page 49 of 225 .3 .1.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 4.1.Dataset tab Figure 21 .

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3

4.1.2 - New Neural Network Wizard
4.1.2.1 - Learning Algorithm tab

Figure 22 - Learning Algorithm tab

Page 50 of 225

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3

4.1.2.2 - Network Settings tab

Figure 23 - Network Settings tab

4.1.2.3 - Dataset tab

Figure 24 - Dataset tab

Page 51 of 225

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3

4.1.3 - About Form

Figure 25 - About form, main view

Page 52 of 225

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3

4.1.3.1 - Datasets View

Figure 26 - About form, Datasets view

Page 53 of 225

Page 54 of 225 .2 .10 -C# User Interface Programming) via events and delegate functions.5 .Working Form Figure 28 .About form.Licenses View Figure 27 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 4.4 .1.Working form 4.Passing Information between Forms Passing the data between the wizard form and the main form takes place as previously described (2.1. Licenses view 4.1.3.

The declaration and association of a delegate and an event takes place in the form that needs to generate an event (9.1.1.6 . lines 266-281).2 -“FrmMain. when instantiated. Then the runworkerasync function executes and then shows the working form.1.cs”.1. 2009). 2007). the processing of the class sent from the Wizard form takes place (see 9.1.1. lines 29-44).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 An event is a method of notifying a listening object that something has occurred. The declaration of a new event handler in the main form takes place prior to displaying the new neural network wizard form (see 9. This leads to a user interface that is much more accessible (Microsoft.1. Firstly. Events. lines 29 – 31). are associated with a delegate function to respond to that event (Lysle.3 -“Wiz1.1.2 -“FrmMain.1.1.cs”.1. (See 9.2 -“FrmMain. The trigger of a new event takes place in the Wizard form when you are on the last tab and the user clicks “next” (see 9.cs” lines 70-119).2 -“FrmMain. lines 306-353.2 “FrmMain.Keeping the User Interface Active whilst Processing is Occurring The user interface delegates the processing to a background worker class.1. In the event handler in the main form. the background worker class is instantiated and supplied with the event handler references functions (see 9.1. 4. The two-event handling functions are at 9.1.cs”.1.cs” lines 120-129).cs” lines 145-163). BackgroundWorker Class. Page 55 of 225 . The “NetworkUpdatedEventArgs” class is a storage class for passing data (see 9.3 -“Wiz1.cs”.1 -“NetworkUpdatedEventHandler”).1.1.

cs” for the C# implementation) The main points to note in the code as it stands are: • The crossover (single-point) and mutation (random increment or decrement) implementations (see 9.1 .2 .cs” lines 176204).5 -“Algorithms\GeneticAlgorithm.cs” lines 32-44).2 -“GANetwork.1. Once this implementation was complete. FANN) which is a C programming language based neural network library. it was both faster than the C++ implementation (due mostly to my use of build in .cs that links into the . I decided to try re-implementing it in C#.2 -“GANetwork.4. • • The sorting implementation in GANetwork. Therefore. The compilation of the FANN library is static.1.NET built in functionality (see 9.1. The neural network specific functionality added to the network model in the network specialisation class GANetwork (see 9.Back-propagation and Cascade Correlation Implementation The FANN library implements both cascade correlation and back propagation (Nissen. and the Dll linked to this to allow access to the functionality. There is an unmanaged C++ wrapper (9.5 “Algorithms\GeneticAlgorithm.2.3.3.1.1.3 Page 56 of 225 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 4. and didn’t suffer from the memory leaks (being managed code) so that’s the version that exists in the application today (see 9. but had issues with memory leaks and was slow.2.cs” lines 51-126) 4.NET sort methods in the sorting part of the genetic algorithm).2.Genetic Algorithm Implementation Although the original design was to implement genetic algorithms in C++. after having used C# extensively in the UI I felt more confident about the performance loss and the use of the programming language itself. when the C++ implementation of genetic algorithms that I wrote functioned.

C# accesses the C++/CLI class directly.Neural Network Implementation Although two previous attempts made at programming a neural network. and nodes) the eventual implementation turned out to be basic yet effective (9. which using the data from the heading (number of epochs of data.1. Three jagged arrays are utilised. the inputs.1.1.h”. 9.4. will also be the coordinates for its corresponding weight (lines 17-19).h”.cpp”. and the outputs. In the constructor (lines 41 – 77).4 .2. which allows it to manipulate the FANN functions in the static library through this class. 9.1. In the initialisation routines (lines 28-124).5 -“UnManaged_FANN_CasC.Data Parser Implementation The data parser (9. via a number of designated separator characters. the initialisation of all the arrays to their correct sizes takes place and then the initialisation of the weights to a random double between zero and one.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 “UnManaged_FANN_BackProp. followed an object-based structure (network. 9.cpp”) for the Cbased FANN functions in the DLL and a C++/CLI wrapper (9. into arrays of strings.cs”).cs”) parses data directly from the rich text boxes that they’re displayed upon on the main form.4. which store respectively.4 “UnManaged_FANN_BackProp. number of inputs.2 -“FANN_Wrapper. the weights. 9.1 -“FANN_Wrapper.h”.4. these arrays form parameters for the other functions.4. From there.1. The inputs and weights are structures so that the coordinates for an input.6 -“UnManaged_FANN_CasC.1.3 . via references to these objects passed down via the hierarchy (references are passed to avoid the necessity of passing entire strings down the hierarchy).3.2 -“DatasetParser.cpp”) for the unmanaged C++ classes. layers.1. 4.1 -“Network. 4.4. number of outputs) separates the data into two-dimensional arrays suitable to be passed to other functions. Page 57 of 225 . the separation of the textual inputs takes place.

so repetition is instead used. All nodes determine their outputs using the sigmoid function (lines 130-133).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 The main functionality of the neural network class can be found in the “internal” region (lines 126-235). As the program presented the data in each case.1 . to run the output layer. The genetic algorithm and back-propagation neural networks each had four hidden nodes for each dataset. 5 .XOR In the XOR dataset. recording took place for analysis. I tested the neural network implementation by creating a neural network with structure 2. 3. 1. Then entered the weights from the neural network program (by means of using the visual studio debugger to check variable values) into the spreadsheet and checked that the outputs coincided. and creating a neural network with a similar structure on an excel spreadsheet.4 -Datasets) (ten times for XOR.Testing Data Testing took place several times for each algorithm and dataset (see 9. 5. The target error was 0. five times for Virus classification and Fishers iris data).01 for XOR and 0. the dataset is too small to split. Page 58 of 225 . and hidden layer which are all called by the public “run()” function (lines 241-270). input layer.02 for Fishers iris data and the virus classification data. there are three functions.

009962559 0.00997699 0.00995716 0.0099765 0.009566951 0. the mean squared error from the network result and the testing mean squared error are different despite the data being the same.009566951 0.00999069 0.0098402 0.009885499 Table 3 .00957518 0.00989974 0.00998959 0.Back-propagation In the back-propagation testing data.00990161 0.2 .5 Result MSqE (GA) 0.009157809 Table 2 .009248076 0.009375132 0.009157809 Testing MSqE (GA) 0.1 .00987211 0.009888967 0.009962559 0.009162436 0.Genetic Algorithm testing data (XOR) 5.00990403 0.009248076 0.009375132 0.00983181 0.1. Iterations (BP) 345 287 319 327 301 323 421 315 362 325 332.00881867 0. because of floating point inaccuracies converting between double precision floating-point numbers in my code and single precision in the FANN code.00988649 0.006150786 0.00957518 0.1.Genetic Algorithms Iterations (GA) 591 407 500 466 499 534 675 557 357 939 552.00988863 0.00992291 0.00991907 0.0099113 0.009162436 0.00881867 0.00998032 0.Back-propagation testing data (XOR) Page 59 of 225 .00992081 0.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 5.5 Result MSqE (BP) 0.00998027 0.009969323 Testing MSqE (BP) 0.00999799 0.009829335 0.009888967 0.006150786 0.009829335 0.

78E-14 1.00798691 0. Nodes (CC) 2 2 2 2 2 2 2 2 2 2 2 Result MSqE (CC) 0.1.43545E-13 Table 4 .00656719 0.3 .00585159 0.83E-13 2.00992534 0.22E-13 1.00855044 0.99E-13 1.00960884 0.18E-13 7.14E-14 2.Cascade Correlation In the cascade correlation data. the call back function triggers.86E-14 1.02E-13 2.00889787 0.00573252 0. At each node addition. which is when the program records the mean squared error.0086449 0.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 5. but in the case of cascade training.00863005 0.32E-13 2.Cascade Correlation testing data (XOR) Page 60 of 225 .008039565 Testing MSqE (CC) 7. the network is trained more after that point (output weight training). the testing result differs quite widely from the network result.02E-13 1.

01 0.Number of iterations to achieve target (XOR) 0.Mean squared error upon last training test (XOR) Page 61 of 225 .004 0.002 0 0 5 10 15 Result MSqE (GA) Result MSqE (BP) Result MSqE (CC) Figure 30 .006 0.008 0.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 5.Graphs 1000 900 800 700 600 500 400 300 200 100 0 0 5 10 15 Iterations (GA) Iterations (BP) Nodes (CC) Figure 29 .012 0.4 .1.

004 0.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 0.006 0.The mean squared error at testing (XOR) Page 62 of 225 .002 0 0 5 10 15 Testing MSqE (GA) Testing MSqE (BP) Testing MSqE (CC) Figure 31 .01 0.012 0.008 0.

0143778 0.018789438 0.0195853 0.Back-propagation testing data (Iris data) 5.0141406 0.019751257 0.0199916 0.0141935 Table 6 .Back-propagation Iterations (BP) 434 529 474 391 527 471 Mean squared error (BP) 0.0199918 0.0164964 0.01950238 Testing mean squared error (CC) 0.1 .014144 0.0199892 0.018789438 0.2.3 .019719802 0.Cascade Correlation Nodes (CC) 2 2 2 2 2 2 Mean squared error (CC) 0.0158679 0.01998996 Testing mean squared error (BP) 0.0199719 0.019751257 0.019763877 0.014116 0.0189456 0.2.019775896 0.0196297 0.Genetic Algorithms Iterations (GA) 1928 1834 1277 2835 8716 3318 Mean squared error (GA) 0.0160925 Table 7 .2 .Cascade correlation testing data (Iris data) Page 63 of 225 .2.0199972 0.01956255614 Table 5 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 5.0198416 0.0193595 0.019763877 0.2 .019719802 0.0199999 0.0140669 0.019775896 0.0139982 0.Fishers Iris data 5.Genetic algorithm testing data (Iris data) 5.01956255614 Testing mean squared error (GA) 0.0143806 0.

0192 0.Graphs 10000 9000 8000 7000 6000 5000 4000 3000 2000 1000 0 0 2 4 6 Iterations (GA) Iterations (BP) Nodes (CC) Figure 32 .Number of iterations to achieve target (Iris data) 0.Mean squared error upon last training test (Iris data) Page 64 of 225 .02 0.0194 0.019 0.0202 0.0196 0.0188 0.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 5.2.4 .0198 0.0186 0 2 4 6 Mean squared error (GA) Mean squared error (BP) Mean squared error (CC) Figure 33.

Virus Classification This dataset covers 61 viruses affecting several crops including tobacco.015 0.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 0.02 Testing mean squared error (GA) Testing mean squared error (BP) Testing mean squared error (CC) 0.3 . There are 18 measurements on each virus. which are the number of amino acid residues per molecule of coat protein.025 0. cucumber. and others. The dataset is in order: • • • • Hordeviruses x3 Tobraviruses x6 Tobamoviruses x39 Furoviruses x13 Page 65 of 225 .The mean squared error at testing (Iris data) 5. tomato.005 0 0 2 4 6 Figure 34.01 0.

Back-propagation Iterations (BP) 3275 10922 2800 3744 6295 5407.0199621 0.2 Mean Squared Error training (BP) 0.00979968 0.2 .0195173 0.0199916 0.8 Mean Squared Error training (GA) 0.1 .019743 0.01983722 Mean Squared Error testing (BP) 0.0194383 0.01870197 0.3.4 Mean Squared Error training (CC) 0.0165979 0.0393106 0.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 5.3.00980208 0.Cascade Correlation Nodes (CC) 1 2 1 2 1 1.016183342 Mean Squared Error testing (CC) 0.0434423 0.019861371 0.03833646 Table 9 .0105845 0.0190612094 Table 8 .019927361 0.0190612094 Mean Squared Error testing (GA) 0.0197935 0.00640291 0.Genetic Algorithms testing data (Viruses) 5.019092293 0.01870197 0.019861371 0.0391179 0.0185152 0.0196959 0.0350318 0.3.019927361 0.0111416 0.011585152 Table 10 .017723052 0.019092293 0.0347797 0.Back-propagation testing data (Viruses) 5.Genetic Algorithms Iterations (GA) 1818 1421 896 11676 1068 3375.017043 0.3 .017723052 0.Cascade Correlation testing data (Viruses) Page 66 of 225 .

02 Mean Squared Error training (GA) Mean Squared Error training (BP) Mean Squared Error training (CC) 0.025 0.Mean squared error upon last training test (Viruses) Page 67 of 225 .005 0 0 2 4 6 Figure 36 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 5.Number of iterations to achieve target (Viruses) 0.3.015 0.Graphs 14000 12000 10000 8000 6000 4000 2000 0 0 2 4 6 Iterations (GA) Iterations (BP) Nodes (CC) Figure 35 .4 .01 0.

1 . genetic algorithms and back-propagation offer similar performance in solving the problem.Comparisons 6. versus backpropagations 333 iterations on average.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 0.035 0.01. The back-propagation implementation seems to be slightly faster in real-time but this is possibly attributable to managed code inefficiencies versus unmanaged code.1.01 0. Page 68 of 225 .005 0 0 2 4 6 Mean Squared Error testing (GA) Mean Squared Error testing (BP) Mean Squared Error testing (CC) Figure 37 . Genetic algorithm solutions are slightly more accurate.03 0.Mean squared error at testing (Viruses) 6 .XOR As regards to the XOR problem.04 0. Genetic algorithms take around five hundred iterations on average to produce a workable solution when aiming for a mean squared error of below 0.05 0.045 0.015 0.Back-propagation and Genetic Algorithms 6.1 .02 0. but to a negligible degree in this scenario. or possible inefficiencies in my programming.025 0.

FANN)) this is around 300 epochs to solve the problem.1.2 . The cascade algorithm also achieves a higher degree of accuracy. The genetic algorithm solutions also held their effectiveness through to the testing data with extreme consistency.Cascade Correlation and Genetic Algorithms 6. With an average of 3318 generations for genetic algorithms versus 471 epochs for back-propagation this became clear. Page 69 of 225 .Fishers Iris Data Genetic algorithms clearly found this a difficult problem to solve as opposed to backpropagation. 6. The back-propagation generated networks were also more accurate overall (in fact on this particular problem analysed the testing data more effectively than the training data).2 .Virus Classification The solution of this problem was more effective by genetic algorithms than by backpropagation.2.Fishers Iris Data Again. 6. on this data set cascade training came to a satisfactory solution faster and more accurately than genetic algorithms. whereas the back-propagation solutions lost effectiveness on the training data. whilst the cascade algorithm used 2 nodes to solve the problems.XOR This problem was solved more effectively by the Cascade algorithm than by genetic algorithms.2 epochs for backpropagation.2 .1.1 . 6. genetic algorithms taking 550 generations on average to solve the problem to a suitable accuracy. with an average of 3375.8 generations versus 5407.3 .2.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 6. At a maximum of 150 epochs per node added (the default (Nissen.

genetic algorithms fall down as a solution. the unavoidable conclusion is that cascade training is the most effective of the three training methods examined.2. Page 70 of 225 . Compared to cascade correlation however.3 .Evaluation Overall.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 6. With more time. with the information presented and discussed here. although there are further investigations I would like to continue with given more time (and may continue with in my own time).Testing conclusions Genetic algorithms are a valuable and effective means of training neural networks. 7 . solutions developed using genetic algorithms tend to hold their effectiveness through to the data set effectively. I feel the project has been a success.Virus Classification With this data set cascade training again came to a satisfactory solution faster and more accurately than genetic algorithms did on the same data. however. They are roughly as effective as back-propagation training and although back-propagation is faster for the most part.3 . 6. more in-depth testing using more varied datasets a clearer comparison and possibly drawbacks associated with cascade training may become evident.

5 -Objectives): Research the fields of genetic algorithms and neural networks. Page 71 of 225 . The only objective that I feel could be improved upon is the final objective. and in which cases these methods are suitable. Identify the most effective ways of training neural networks using genetic algorithms. Build a user-interface for the aforementioned application allowing for easy alteration of appropriate variables. “Identify the most effective ways of training neural networks using genetic algorithms. and that allows manual adjustment of relevant variables. as well as previous research. Research current information on training neural networks using genetic algorithms. Using above application. leading to a more solid conclusion and more information on the various strengths and weaknesses of the various learning algorithms examined.How many of the original objectives were achieved? The original objectives were as follows (9. and in which cases these methods are suitable.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 7.” With more time more advanced tests could be undertaken using larger data sets.1 . Build an application that is capable of training a neural network with a genetic algorithm and with a backpropagation system. evaluate training methods for neural networks.

Testing both with and without scaling and utilising different scaling methods. the program occasionally causes an exception (most specifically when creating cascade or back-propagation networks in quick succession). These could be resolved fairly simply (I believe) by implementing the iDispose interface in my managed code to allow it to be destroyed upon demand instead of waiting for the garbage collector.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 8 .Possible Improvements More advanced testing would be the biggest improvement that more time would make possible. Improve memory management. Due to small conflicts between managed and unmanaged memory. Testing with datasets that would take large amounts of time to train. Some forms of testing that would be possible are: • • • • • Testing with datasets with larger numbers of inputs Testing with large datasets that attempt to train the network to distinguish between small differences in data. Page 72 of 225 .

Implement the back-propagation algorithm in C# instead of C++.Possible Training Algorithm Improvements With more time. allowing more to be undertaken successfully. (the new network wizard and the working form) lock out the rest of the program. as well as implementing roulette wheel selection. Having the back-propagation algorithm in C# similar to the genetic algorithm would also allow a more “apples to apples” comparison. experimenting with multiple point crossover and other crossover forms. but still minimize instead of blocking other computer programs. I would experiment with the following training algorithm improvements: • Improve the genetic algorithms crossover and mutation algorithms.Possible User Interface Improvements The main improvements to the user interface that I would make if I had more time are: • • Increased number of variable options presented to the user (with appropriate defaults set).1 . • Implement the cascade training algorithm in C# instead of C++. Page 73 of 225 . 8.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 8. Contrary to my opinion before undertaking extensive usage of C# I now believe the reduced development time.2 . Making the forms shown as dialogs. • • Attempt to improve the speed of the genetic algorithm. for the reasons stated above. is worth the possibility of reduced performance.

Although there was another section of the program compiled from source (the FANN library). which this code is partially based upon. and neither is the code for the prototype constructed in milestone two.1 . so that this listing contains only code I have personally written for the project.Source code The source code in this section is organised first by which part of the application it represents and then by what source file it originally resided in. Page 74 of 225 . I have not included it in the source code listing. This section also contains only the final draft of the code.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 9 . The earlier attempts at genetic algorithm and neural network implementations are not included.Appendices 9. The code contains line numbering for pinpointing the precise lines in question within the body-text when crossreferencing.

William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 9. System. private bool _CC_ITerr.EventArgs { #region data /********************\ |* DATA *| \********************/ //Training Algorithm Identifier private int trainingalg. namespace _05025397.1.1 .Text. private double _CC_LearningRate. System.Classes { public class NetworkUpdateEventArgs : System. private int _CC_Reports.Linq.Collections.“NetworkUpdatedEventHandler” 00000001 00000002 00000003 00000004 00000005 00000006 00000007 00000008 00000009 00000010 00000011 00000012 00000013 00000014 00000015 00000016 00000017 00000018 00000019 00000020 00000021 00000022 00000023 00000024 00000025 00000026 using using using using System.Controller. //Cascade Correlation private double _CC_Terr.1. Page 75 of 225 .1. private int _CC_MaxNeurons. System. //Dataset identifier private int _TrDataSet.View 9.1 .Generic.

#endregion #region constructor /********************\ |* CONSTRUCTOR *| \********************/ //Constructor to accept all necessary data public NetworkUpdateEventArgs (int Algorithm. int CCMaxNeurons. private int _BP_HiddenL. Page 76 of 225 . bool CCITerr. private int _GA_Crossover. private int _GA_Mutation. int GAPopSize. private int _GA_GenLimit. //BackProp private int _BP_EpochLimit. double CCLearningRate. double CCTerr. private double _BP_LearningRate. private int _BP_Reports.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000027 00000028 00000029 00000030 00000031 00000032 00000033 00000034 00000035 00000036 00000037 00000038 00000039 00000040 00000041 00000042 00000043 00000044 00000045 00000046 00000047 00000048 00000049 00000050 00000051 00000052 00000053 00000054 00000055 00000056 //Genetic Algorithm private int _GA_PopSize. private double _GA_Terr. private double _BP_Terr. private bool _GA_ITerr. private bool _BP_ITerr. int CCReports. private int _GA_Reports. private int _GA_HiddenL.

_GA_Mutation = GAMutation. int BPHiddenL. _GA_ITerr = GAITerr. _CC_MaxNeurons = CCMaxNeurons. double BPTerr. //Cascade Correlation _CC_Terr = CCTerr. int GAMutation. Page 77 of 225 . _CC_ITerr = CCITerr. int BPEpochLimit. _GA_Reports = GAReports. int GACrossOver. _GA_Crossover = GACrossOver. //Genetic Algorithm _GA_PopSize = GAPopSize. _CC_LearningRate = CCLearningRate. _GA_Terr = GATerr. _CC_Reports = CCReports. int BPReports. double GATerr.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000057 00000058 00000059 00000060 00000061 00000062 00000063 00000064 00000065 00000066 00000067 00000068 00000069 00000070 00000071 00000072 00000073 00000074 00000075 00000076 00000077 00000078 00000079 00000080 00000081 00000082 00000083 00000084 00000085 00000086 int GAGenLimit. int TrData_Set) { //Algorithm trainingalg = Algorithm. bool GAITerr. int GAReports. //Backpropagation Algorithm _BP_EpochLimit = BPEpochLimit. int GAHiddenL. _GA_HiddenL = GAHiddenL. _GA_GenLimit = GAGenLimit. double BPLearningRate. bool BPITerr.

_TrDataSet = TrData_Set. } } #endregion #region genetic_algorithms //Genetic Algorithm public int GA_PopSize { get { return _GA_PopSize. } } public int CC_Reports { get { return _CC_Reports. } } public bool CC_ITerr { get { return _CC_ITerr.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000087 00000088 00000089 00000090 00000091 00000092 00000093 00000094 00000095 00000096 00000097 00000098 00000099 00000100 00000101 00000102 00000103 00000104 00000105 00000106 00000107 00000108 00000109 00000110 00000111 00000112 00000113 00000114 00000115 00000116 _BP_HiddenL = BPHiddenL. } } public int GA_GenLimit Page 78 of 225 . _BP_Terr = BPTerr. _BP_Reports = BPReports. _BP_ITerr = BPITerr. _BP_LearningRate = BPLearningRate. } } public double CC_LearningRate { get { return _CC_LearningRate. } } public int CC_MaxNeurons { get { return _CC_MaxNeurons. } #endregion #region getters/setters #region cascade_correlation //Cascade Correlation public double CC_Terr { get { return _CC_Terr.

} } #endregion #region back_propagation //Back propagation public int BP_EpochLimit { get { return _BP_EpochLimit. } } public int GA_Crossover { get { return _GA_Crossover. } } public int GA_Reports { get { return _GA_Reports. } } public double BP_Terr { get { return _BP_Terr. } } public bool GA_ITerr { get { return _GA_ITerr. } } public int BP_Reports { get { return _BP_Reports. } } public bool BP_ITerr { get { return _BP_ITerr. } } public int GA_Mutation { get { return _GA_Mutation. } } public int GA_HiddenL { get { return _GA_HiddenL. } } #endregion Page 79 of 225 . } } public int BP_HiddenL { get { return _BP_HiddenL. } } public double GA_Terr { get { return _GA_Terr.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000117 00000118 00000119 00000120 00000121 00000122 00000123 00000124 00000125 00000126 00000127 00000128 00000129 00000130 00000131 00000132 00000133 00000134 00000135 00000136 00000137 00000138 00000139 00000140 00000141 00000142 00000143 00000144 00000145 00000146 { get { return _GA_GenLimit. } } public double BP_LearningRate { get { return _BP_LearningRate.

Linq.2 . System.1.Forms. System.Drawing. } } #endregion } 9. namespace _05025397. System.“FrmMain.Windows. System.Text.IO. using _05025397.Data. } } public int TrainingAlgorithm { get { return trainingalg. System.Collections.Controller { public partial class FrmMain : Form Page 80 of 225 . System. using ZedGraph.ComponentModel. System.1. System.cs” 00000001 00000002 00000003 00000004 00000005 00000006 00000007 00000008 00000009 00000010 00000011 00000012 00000013 00000014 00000015 00000016 00000017 using using using using using using using using using System.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000147 00000148 00000149 00000150 00000151 00000152 00000153 00000154 00000155 00000156 00000157 } #endregion #region external //General public int TrDataSet { get { return _TrDataSet.Generic.

RunWorkerCompleted += new RunWorkerCompletedEventHandler(bw_RunWorkerCompleted).icon. //Working dialog Controller.Algorithms. //Background thread static BackgroundWorker bw = new BackgroundWorker(). this. #endregion #region constructor /********************\ |* CONSTRUCTOR *| \********************/ public FrmMain() { InitializeComponent().Icon = Properties.Working workingdialog = new Controller.Resources. bw.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000018 00000019 00000020 00000021 00000022 00000023 00000024 00000025 00000026 00000027 00000028 00000029 00000030 00000031 00000032 00000033 00000034 00000035 00000036 00000037 00000038 00000039 00000040 00000041 00000042 00000043 00000044 00000045 00000046 00000047 { #region data /********************\ |* DATA *| \********************/ //The neuralnetwork Controller. } #endregion #region internal_functions /********************\ Page 81 of 225 .DoWork += new DoWorkEventHandler(bw_DoWork). bw.TrainingAlgorithm network.Working().

Text = "Iteration".TrainingAlgorithm == 0) { Page 82 of 225 .zedGraphControl1. MSE. // Hide the legend MSEpane. this. Classes.zedGraphControl1.zedGraphControl1.TrDataSet).Text = "Mean Squared Error".CurveList.Circle).GraphPane. Color.Title. GraphPane MSEpane = this. } private void updatenetwork(object sender.Red.AddCurve("".Legend. MSEpane. MSEpane.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000048 00000049 00000050 00000051 00000052 00000053 00000054 00000055 00000056 00000057 00000058 00000059 00000060 00000061 00000062 00000063 00000064 00000065 00000066 00000067 00000068 00000069 00000070 00000071 00000072 00000073 00000074 00000075 00000076 00000077 |* INTERNAL *| \********************/ private void plotgraphsMSE(PointPairList MSE) { this.GraphPane.NetworkUpdateEventArgs e) { LoadDataset(e.YAxis.IsVisible = false. //Titles MSEpane. SymbolType.Title.Title.XAxis.Text = "MSE".AxisChange().Clear(). MSEpane. try { if (e.

GA_ITerr. e. txtTestData). txtTrainData.BP_LearningRate. e.BackProp (e. } } catch { Page 83 of 225 . e.CC_Reports.BP_HiddenL.BP_Terr. txtTrainData. e. e.CC_Terr.GA_PopSize.CascadeCorrelation (e. e.BP_EpochLimit. txtTestData).CC_MaxNeurons. txtTrainData. e.BP_Reports.GA_Reports.BP_ITerr.GA_Terr.Algorithms.GA_HiddenL.CC_ITerr.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000078 00000079 00000080 00000081 00000082 00000083 00000084 00000085 00000086 00000087 00000088 00000089 00000090 00000091 00000092 00000093 00000094 00000095 00000096 00000097 00000098 00000099 00000100 00000101 00000102 00000103 00000104 00000105 00000106 00000107 network = new Controller.TrainingAlgorithm == 2) { network = new Controller.GA_GenLimit. e.GA_Crossover.GA_Mutation. } if (e. e. e.GeneticAlgorithm (e. e.TrainingAlgorithm == 1) { network = new Controller.CC_LearningRate. e. e. txtTestData).Algorithms.Algorithms. e. e. e. } if (e.

} private void LoadDataset(int Data_set) { switch (Data_set) { case (int)Dataset.Text = Properties. } } private void new_network() { Controller.Wiz1.Text = network.Wiz1().OK.ShowDialog().Show("Error creating new network".network_details(). wizard.Text = Properties.Resources.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000108 00000109 00000110 00000111 00000112 00000113 00000114 00000115 00000116 00000117 00000118 00000119 00000120 00000121 00000122 00000123 00000124 00000125 00000126 00000127 00000128 00000129 00000130 00000131 00000132 00000133 00000134 00000135 00000136 00000137 MessageBox.Exit(). Page 84 of 225 . wizard. } if (network != null) { txtNetSettings. "Error". txtTestData.XOR: txtTrainData.NetworkUpdateHandler(updatenetwork).XORtest. Application.Resources.Wiz1 wizard = new Controller.XORtrain. MessageBoxIcon.NetworkUpdated += new Controller.Error). MessageBoxButtons.

Resources. case (int)Dataset.FISHERtrain. try { train = LoadTxtFile(trainpath).FISHER: txtTrainData.Text = Properties.Show("Error: Problem loading training data".Text = Properties.OK.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000138 00000139 00000140 00000141 00000142 00000143 00000144 00000145 00000146 00000147 00000148 00000149 00000150 00000151 00000152 00000153 00000154 00000155 00000156 00000157 00000158 00000159 00000160 00000161 00000162 00000163 00000164 00000165 00000166 00000167 break. break.\\". } Page 85 of 225 . string test = "". string train = "". } } private void LoadCustomDataSet() { string trainpath = SelectTextFile(".\\".FISHERtest. break. "Error!".Error). } catch { MessageBox. string testpath = SelectTextFile(". "Select Training Data"). "Select Testing Data").CUSTOM: LoadCustomDataSet(). txtTestData. MessageBoxIcon. case (int)Dataset.Resources. MessageBoxButtons.

FileAccess. } catch { MessageBox. Page 86 of 225 .Open. Reader = new StreamReader(File). "Error!".OK.Error). } private string LoadTxtFile(string path) { string data = "".Error). MessageBoxIcon. } txtTestData.OK. data = Reader. } catch { MessageBox. MessageBoxIcon. FileMode.ReadToEnd(). MessageBoxButtons. txtTrainData. MessageBoxButtons.Read).Text = train. FileStream File = null. try { File = new FileStream(path.Text = test. "Error!".Show("Error: Problem loading testing data".William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000168 00000169 00000170 00000171 00000172 00000173 00000174 00000175 00000176 00000177 00000178 00000179 00000180 00000181 00000182 00000183 00000184 00000185 00000186 00000187 00000188 00000189 00000190 00000191 00000192 00000193 00000194 00000195 00000196 00000197 try { test = LoadTxtFile(testpath).Show("Error reading selected file: " + path. StreamReader Reader = null.

Filter = "txt files (*. dialog.ShowDialog() == DialogResult. if (File != null) File. dialog. return (dialog.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000198 00000199 00000200 00000201 00000202 00000203 00000204 00000205 00000206 00000207 00000208 00000209 00000210 00000211 00000212 00000213 00000214 00000215 00000216 00000217 00000218 00000219 00000220 00000221 00000222 00000223 00000224 00000225 00000226 00000227 data = "".txt|All files (*. dialog.*)|*. } private string SelectTextFile(string initialDirectory.Title = title.Close().*".OK) ? dialog. } return data. } #endregion #region menu_items /********************\ |* MENU ITEMS *| \********************/ Page 87 of 225 .txt)|*. } finally { if (Reader != null) Reader.InitialDirectory = initialDirectory. string title) { OpenFileDialog dialog = new OpenFileDialog().FileName : null.Close().

EventArgs e) { FrmAbout About = new FrmAbout(). EventArgs e) { if (network != null) { if (network. EventArgs e) { new_network().Trained != false) { network.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000228 00000229 00000230 00000231 00000232 00000233 00000234 00000235 00000236 00000237 00000238 00000239 00000240 00000241 00000242 00000243 00000244 00000245 00000246 00000247 00000248 00000249 00000250 00000251 00000252 00000253 00000254 00000255 00000256 00000257 private void newNetworkToolStripMenuItem_Click(object sender. EventArgs e) { this.Show().Close(). } private void exitToolStripMenuItem_Click(object sender. } #endregion #region buttons Page 88 of 225 .Test(). } private void runToolStripMenuItem_Click(object sender. } } } private void aboutToolStripMenuItem1_Click(object sender. About.

Trained) Page 89 of 225 . if (!bw. EventArgs e) { //Shouldn't be busy if the user managed to click this //but just make 100% certain.RunWorkerAsync(). } } } } private void btnTest_Click(object sender.ShowDialog().IsBusy) { if (network != null) { if (!network. EventArgs e) { if (network != null) { if (network.Trained) { bw.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000258 00000259 00000260 00000261 00000262 00000263 00000264 00000265 00000266 00000267 00000268 00000269 00000270 00000271 00000272 00000273 00000274 00000275 00000276 00000277 00000278 00000279 00000280 00000281 00000282 00000283 00000284 00000285 00000286 00000287 /********************\ |* BUTTONS *| \********************/ private void btnNew_Click(object sender. workingdialog. } private void btnTrain_Click(object sender. EventArgs e) { new_network().

Test().ReportData. } } } } } #endregion #region bg_worker /********************\ |* BG WORKER *| \********************/ private void bw_DoWork(object sender. DoWorkEventArgs e) { bool success = true.Length. txtOutput. if (network != null) { success = network. if (success) { txtOutput.Tested) { bool success = network.Text.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000288 00000289 00000290 00000291 00000292 00000293 00000294 00000295 00000296 00000297 00000298 00000299 00000300 00000301 00000302 00000303 00000304 00000305 00000306 00000307 00000308 00000309 00000310 00000311 00000312 00000313 00000314 00000315 00000316 00000317 { if (!network. txtOutput. } Page 90 of 225 .SelectionStart = txtOutput.Text = network.ScrollToCaret(). txtOutput.Clear().Refresh().Train(). txtOutput.

Result = success.Text.Result.getGraphData()). MessageBoxIcon.Show("Error during network train: " + e. } else { MessageBox.Show("Error training network!".ScrollToCaret().ToString(). } private void bw_RunWorkerCompleted(object sender.Text = network. "Error".Error).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000318 00000319 00000320 00000321 00000322 00000323 00000324 00000325 00000326 00000327 00000328 00000329 00000330 00000331 00000332 00000333 00000334 00000335 00000336 00000337 00000338 00000339 00000340 00000341 00000342 00000343 00000344 00000345 00000346 00000347 e. txtOutput. txtOutput. txtOutput. plotgraphsMSE(network. MessageBoxIcon.Error != null) { MessageBox.Length.Clear(). MessageBoxButtons.OK.Error).SelectionStart = txtOutput. "Error training network!". //Check for errors if (e. if (success) { txtOutput. RunWorkerCompletedEventArgs e) { bool success. MessageBoxButtons.Error.ReportData. } success = (bool) e.OK. } Page 91 of 225 .

Drawing.ComponentModel.Linq. System.Collections. Page 92 of 225 .Hide().Data.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000348 00000349 00000350 00000351 00000352 00000353 00000354 00000355 00000356 00000357 00000358 00000359 00000360 00000361 00000362 00000363 00000364 00000365 00000366 } //Remove the dialog locking out the main form //and showing that we're currently working workingdialog. System.1.“Wiz1.3 .1.cs” 00000001 00000002 00000003 00000004 00000005 00000006 using using using using using using System. } #endregion #region misc /********************\ |* MISC *| \********************/ private void FrmMain_Shown(object sender. System.Generic. } #endregion } 9. System. EventArgs e) { //Run the new neural net wizard new_network(). System.

NetworkUpdateEventArgs e). #endregion public partial class Wiz1 : Form { #region data /********************\ |* DATA *| \********************/ public delegate void NetworkUpdateHandler(object sender. namespace _05025397. private int algorithm. Page 93 of 225 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000007 00000008 00000009 00000010 00000011 00000012 00000013 00000014 00000015 00000016 00000017 00000018 00000019 00000020 00000021 00000022 00000023 00000024 00000025 00000026 00000027 00000028 00000029 00000030 00000031 00000032 00000033 00000034 00000035 00000036 using System. Classes. CUSTOM }. using _05025397.Controller { #region enums /********************\ |* ENUMS *| \********************/ enum Training { GA. private int Trainset.Windows. BP. CC }. FISHER.Text.Forms. using System. public event NetworkUpdateHandler NetworkUpdated. enum Dataset { XOR.

Trainset = (int)Dataset. EventArgs e) Page 94 of 225 .Close(). EventArgs e) { this.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000037 00000038 00000039 00000040 00000041 00000042 00000043 00000044 00000045 00000046 00000047 00000048 00000049 00000050 00000051 00000052 00000053 00000054 00000055 00000056 00000057 00000058 00000059 00000060 00000061 00000062 00000063 00000064 00000065 00000066 #endregion data #region constructor /********************\ |* CONSTRUCTOR *| \********************/ public Wiz1() { InitializeComponent().icon. algorithm = (int)Training.XOR. } #endregion #region control_events /********************\ |* CONTROL EVENTS *| \********************/ #region buttons /********************\ |* BUTTONS *| \********************/ private void btnCancel_Click(object sender. } private void rBtnBackprop_CheckedChanged(object sender.Icon = Properties.Resources. this.GA.

Image = _05025397.Visible = false. panCC. Page 95 of 225 . } } private void rBtnCascadeCorrelation_CheckedChanged(object sender.Visible = false.Checked == true) { picAlg. lblCC.Visible = true.Properties.Visible = false.Resources. EventArgs e) { if (rBtnCascadeCorrelation. panGeneticAlgorithm.CC.BP.Properties. panBackprop. lblCC.Visible = true. algorithm = (int)Training.Visible = false. panGeneticAlgorithm.IMAGECascadeCorrelation.Visible = true. panBackprop.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000067 00000068 00000069 00000070 00000071 00000072 00000073 00000074 00000075 00000076 00000077 00000078 00000079 00000080 00000081 00000082 00000083 00000084 00000085 00000086 00000087 00000088 00000089 00000090 00000091 00000092 00000093 00000094 00000095 00000096 { if (rBtnBackprop. lblGA.IMAGEBackProp.Visible = false.Visible = true.Visible = false. lblBP. panCC. lblBP.Visible = false. lblGA.Checked == true) { picAlg.Image = _05025397.Visible = false. algorithm = (int)Training.Resources.

EventArgs e) { if (rBtnGeneticAlgorithm. EventArgs e) { if (tabControl1.Visible = false.Visible = false.Visible = true.Resources. } } private void btnBack_Click(object sender. lblBP.1).Visible = false.SelectedIndex .GA.IMAGEGeneticAlgorithm.Image = _05025397. panCC.Visible = true. lblGA.Properties.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000097 00000098 00000099 00000100 00000101 00000102 00000103 00000104 00000105 00000106 00000107 00000108 00000109 00000110 00000111 00000112 00000113 00000114 00000115 00000116 00000117 00000118 00000119 00000120 00000121 00000122 00000123 00000124 00000125 00000126 } } private void rBtnGeneticAlgorithm_CheckedChanged(object sender. panGeneticAlgorithm. lblCC. } } private void btnNext_Click(object sender.Visible = false. panBackprop. algorithm = (int)Training. EventArgs e) { Page 96 of 225 .Checked == true) { picAlg.SelectedIndex > 0) { tabControl1.SelectTab(tabControl1.

GA_Reports.Text. txtBP_Report. check.Checked. txtBP_EpochLimit.CC_LearningRate. if (check.CC_ITerr.SelectTab(tabControl1.1) { tabControl1.Text. check.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000127 00000128 00000129 00000130 00000131 00000132 00000133 00000134 00000135 00000136 00000137 00000138 00000139 00000140 00000141 00000142 00000143 00000144 00000145 00000146 00000147 00000148 00000149 00000150 00000151 00000152 00000153 00000154 00000155 00000156 if (tabControl1. chkGA_IgnoreTarget.Text.NetworkUpdateEventArgs args = new Classes.CC_Terr.Text.Checked. txtGA_ProbX. txtCC_Report. } else { VerifyData check = new VerifyData (algorithm. check.Text. chkBP_IgnoreTarget. txtGA_Report. check.GA_Mutation.CC_MaxNeurons. txtBP_LearnRate.GA_GenLimit. Page 97 of 225 . check.SelectedIndex < tabControl1. check. txtCC_TargetError. txtBP_Terr. check.Text.Text. check.BP_HiddenL.GA_Terr.NetworkUpdateEventArgs (check. CC_ITerr. check.GA_Crossover.BP_EpochLimit.Checked.GA_ITerr.GA_PopSize. txtCC_MaxNeurons. Trainset).CC_Reports. check. check. check.Text.Text.Text.Text. check.SelectedIndex + 1). txtCC_LearningRate. check. check.Verified) { Classes. txtBP_HiddenNodes. txtGA_ProbM.TabCount .Text. txtGA_HiddenNodes. txtGA_MaxGen.Text.GA_HiddenL.Text. txtGA_PopSize.Text.Text.TrainingAlgorithm. txtGA_Terr.

BP_Terr. } else { MessageBox.BP_Reports.Image = _05025397.FISHER. check.Checked) { picDataset. Trainset = (int)Dataset. args).TrDataSet). } } private void chkDatasetXOR_CheckedChanged(object sender. NetworkUpdated(this.BP_LearningRate. EventArgs e) { if (chkDatasetFisher.Exclamation).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000157 00000158 00000159 00000160 00000161 00000162 00000163 00000164 00000165 00000166 00000167 00000168 00000169 00000170 00000171 00000172 00000173 00000174 00000175 00000176 00000177 00000178 00000179 00000180 00000181 00000182 00000183 00000184 00000185 00000186 check.Dispose(). this.IMAGEIrisFlowers. check. } } } #endregion #region checkboxes /********************\ |* CHECKBOXES *| \********************/ private void chkDatasetFisher_CheckedChanged(object sender.Resources.Properties.Show("Error Checking Data: Data could not be verified". check. MessageBoxButtons.OK. "Error Checking Data".BP_ITerr. MessageBoxIcon. EventArgs e) Page 98 of 225 . check.

IMAGEQuestionMark. EventArgs e) { if (tabControl1. Trainset = (int)Dataset.IMAGEXOR. EventArgs e) { if (chkDatasetCustom. } Page 99 of 225 .Properties.SelectedIndex == 1) { if (panCC. } } private void chkDatasetCustom_CheckedChanged(object sender.Properties. Trainset = (int)Dataset.Checked) { picDataset.Image = _05025397.Resources.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000187 00000188 00000189 00000190 00000191 00000192 00000193 00000194 00000195 00000196 00000197 00000198 00000199 00000200 00000201 00000202 00000203 00000204 00000205 00000206 00000207 00000208 00000209 00000210 00000211 00000212 00000213 00000214 00000215 00000216 { if (chkDatasetXOR.CUSTOM.Resources.Checked) { picDataset.Focus().XOR. } } #endregion #region misc /********************\ |* MISC *| \********************/ private void tabControl1_SelectedIndexChanged(object sender.Visible == true) { txtCC_TargetError.Image = _05025397.

icon.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000217 00000218 00000219 00000220 00000221 00000222 00000223 00000224 00000225 00000226 00000227 00000228 00000229 00000230 00000231 00000232 00000233 00000234 00000235 00000236 } if (panBackprop.Icon = Properties.SelectTab(0).Visible == true) { txtBP_EpochLimit.Visible == true) { txtGA_PopSize. tabControl1. } } } private void Wiz1_Load(object sender. } if (panGeneticAlgorithm.Resources. EventArgs e) { this. } #endregion #endregion } Page 100 of 225 .Focus().Focus().

System. namespace _05025397.Collections.Generic.Drawing.Data.Linq.Text.ComponentModel. EventArgs e) { Page 101 of 225 . System.Controller { public partial class FrmAbout : Form { #region constructor /********************\ |* CONSTRUCTOR *| \********************/ public FrmAbout() { InitializeComponent().4 .1. System. } #endregion #region form /********************\ |* FORM *| \********************/ private void FrmAbout_Load(object sender.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 9.“FrmAbout. System. System.cs” 00000001 00000002 00000003 00000004 00000005 00000006 00000007 00000008 00000009 00000010 00000011 00000012 00000013 00000014 00000015 00000016 00000017 00000018 00000019 00000020 00000021 00000022 00000023 00000024 00000025 00000026 00000027 00000028 00000029 using using using using using using using using System.Forms.Windows.1. System. System.

LICENCE_LGPL21. textBox5.Text).Information).Icon = Properties. LinkLabelLinkClickedEventArgs e) Page 102 of 225 . MessageBoxIcon. label5. MessageBoxButtons.Process.OK.".Resources.Diagnostics. LinkLabelLinkClickedEventArgs e) { // Specify that the link was visited.ProductVersion.LinkVisited = true.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000030 00000031 00000032 00000033 00000034 00000035 00000036 00000037 00000038 00000039 00000040 00000041 00000042 00000043 00000044 00000045 00000046 00000047 00000048 00000049 00000050 00000051 00000052 00000053 00000054 00000055 00000056 00000057 00000058 00000059 this.linkLabel1.Text = Properties.icon.Resources. try { // Navigate to a URL. "Undefined Application".Text = Application. System. this. } } private void linkLabel4_LinkClicked(object sender.Show("No e-mail program defined. } catch { MessageBox. } #endregion #region linklabels /********************\ |* LINKLABELS *| \********************/ private void linkLabel1_LinkClicked(object sender.Start("mailto:" + linkLabel1.

Diagnostics. } } private void linkLabel5_LinkClicked(object sender.Text).Error).Process. "Error".Process.Start (linkLabel5. MessageBoxIcon.Error). MessageBoxIcon. LinkLabelLinkClickedEventArgs e) { try { System.Text). MessageBoxButtons.Show("Error opening link.OK. } catch { MessageBox. MessageBoxButtons.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000060 00000061 00000062 00000063 00000064 00000065 00000066 00000067 00000068 00000069 00000070 00000071 00000072 00000073 00000074 00000075 00000076 00000077 00000078 00000079 00000080 00000081 00000082 00000083 00000084 00000085 00000086 00000087 00000088 00000089 { try { System.Show("Error opening link.". } catch { MessageBox.OK. } } private void linkLabel6_LinkClicked(object sender.". "Error". LinkLabelLinkClickedEventArgs e) Page 103 of 225 .Start (linkLabel4.Diagnostics.

".Start ("http://www. } } private void linkLabel2_LinkClicked(object sender. LinkLabelLinkClickedEventArgs e) { try { System. MessageBoxIcon. MessageBoxButtons.Error).Text). } catch { MessageBox. "Error".snikks.OK.Process.co.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000090 00000091 00000092 00000093 00000094 00000095 00000096 00000097 00000098 00000099 00000100 00000101 00000102 00000103 00000104 00000105 00000106 00000107 00000108 00000109 00000110 00000111 00000112 00000113 00000114 00000115 00000116 00000117 00000118 00000119 { try { System. } catch { MessageBox. MessageBoxButtons.Show("Error opening link.Start (linkLabel6.Error).Show("Error opening link.uk"). } } private void linkLabel7_LinkClicked(object sender.Diagnostics.".OK. MessageBoxIcon.Diagnostics.Process. LinkLabelLinkClickedEventArgs e) Page 104 of 225 . "Error".

org/licenses/old-licenses/lgpl-2.Visible != true) { groupBox1.". groupBox2.html").Visible = false. } catch { MessageBox.Process. "Error".Visible = false.gnu. } else Page 105 of 225 .1.Show("Error opening link.Start ("http://www. pictureBox1.Diagnostics.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000120 00000121 00000122 00000123 00000124 00000125 00000126 00000127 00000128 00000129 00000130 00000131 00000132 00000133 00000134 00000135 00000136 00000137 00000138 00000139 00000140 00000141 00000142 00000143 00000144 00000145 00000146 00000147 00000148 00000149 { try { System. } } #endregion #region buttons /********************\ |* BUTTONS *| \********************/ private void btnDatasets_Click(object sender. tabControl1.Error). MessageBoxButtons.Visible = true.Visible = false. EventArgs e) { if (tabControl1. MessageBoxIcon.OK.

tabControl1. EventArgs e) { if (groupBox2.Visible = false.Visible pictureBox1.Visible = false. pictureBox1.Visible = true. = false.Visible = true.Visible } } private void btnLicences_Click(object sender.Visible = false. EventArgs e) { this.Visible = groupBox2. tabControl1. groupBox1.Visible != true) { groupBox2. groupBox2. false.Visible = true. } else { groupBox1.Dispose(). pictureBox1. } true.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000150 00000151 00000152 00000153 00000154 00000155 00000156 00000157 00000158 00000159 00000160 00000161 00000162 00000163 00000164 00000165 00000166 00000167 00000168 00000169 00000170 00000171 00000172 00000173 00000174 00000175 00000176 00000177 00000178 00000179 { groupBox1.Visible = false. } } private void btnOkay_Click(object sender. Page 106 of 225 .Visible = false.Visible = tabControl1. = true.

EventArgs e) { if (radioButton1.Resources.Checked == true) textBox5.Resources.LICENCE_GPL30. } #endregion } Page 107 of 225 . EventArgs e) { if (radioButton2.LICENCE_LGPL30.Text = Properties.Checked == true) textBox5.Resources.Checked == true) textBox5.Text = Properties.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000180 00000181 00000182 00000183 00000184 00000185 00000186 00000187 00000188 00000189 00000190 00000191 00000192 00000193 00000194 00000195 00000196 00000197 00000198 00000199 00000200 00000201 00000202 00000203 00000204 00000205 } #endregion #region radiobuttons /********************\ |* RADIOBUTTONS *| \********************/ private void radioButton1_CheckedChanged(object sender.Text = Properties. } private void radioButton2_CheckedChanged(object sender. } private void radioButton3_CheckedChanged(object sender.LICENCE_LGPL21. EventArgs e) { if (radioButton3.

} #endregion } Page 108 of 225 .Forms.Icon = Properties. System.Data.Beep. System.SystemSounds.Drawing.Controller { public partial class Working : Form { #region functions /********************\ |* FUNCTIONS *| \********************/ public Working() { InitializeComponent().1.Linq.Collections.icon.Resources.Generic. EventArgs e) { System.5 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 9. System. System.Windows. System. System.Media. } private void Working_Click(object sender. namespace _05025397.cs” 00000001 00000002 00000003 00000004 00000005 00000006 00000007 00000008 00000009 00000010 00000011 00000012 00000013 00000014 00000015 00000016 00000017 00000018 00000019 00000020 00000021 00000022 00000023 00000024 00000025 00000026 00000027 00000028 00000029 using using using using using using using using System. this.1.Play().“Working.Text. System.ComponentModel.

//Training Algorithm Identifier //1 = genetic algorith //2 = backprop //3 = cascade correlation private int trainingalg.2 .Collections.cs” 00000001 00000002 00000003 00000004 00000005 00000006 00000007 00000008 00000009 00000010 00000011 00000012 00000013 00000014 00000015 00000016 00000017 00000018 00000019 00000020 00000021 00000022 00000023 00000024 00000025 using using using using using System.1.Text. System.Windows. System. System. System.“VerifyData. namespace _05025397.Forms.Controller { public class VerifyData { #region data /********************\ |* DATA *| \********************/ //Dataverified variable private bool verified.Controller 9.1 .Generic.1.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000030 } 9.Linq. //Dataset identifer //0 = XOR Page 109 of 225 .2.

//BackProp private int _BP_EpochLimit. //Cascade Correlation private double _CC_Terr. private double _CC_LearningRate. private int _GA_Mutation. private int _GA_GenLimit. private int _BP_Reports. private int _GA_HiddenL.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000026 00000027 00000028 00000029 00000030 00000031 00000032 00000033 00000034 00000035 00000036 00000037 00000038 00000039 00000040 00000041 00000042 00000043 00000044 00000045 00000046 00000047 00000048 00000049 00000050 00000051 00000052 00000053 00000054 00000055 //1 = FISHERS IRIS DATA //2 = CUSTOM private int _TDataSet. private double _GA_Terr. private double _BP_Terr. //Genetic Algorithm private int _GA_PopSize. private bool _GA_ITerr. private bool _CC_ITerr. #endregion Page 110 of 225 . private int _GA_Crossover. private double _BP_LearningRate. private int _CC_MaxNeurons. private int _CC_Reports. private bool _BP_ITerr. private int _GA_Reports. private int _BP_HiddenL.

string CCMaxNeurons. if (_TDataSet != (int)Dataset. string BPEpochLimit. string GAReports. bool CCITerr. string GAMutation. //Algorithm trainingalg = Algorithm. bool BPITerr. string BPLearningRate. string GACrossOver. string GATerr. int TData_Set) { //Initially set to true. string GAGenLimit. //DataSet _TDataSet = TData_Set. string CCTerr. bool GAITerr. string BPHiddenL. string BPTerr. string GAHiddenL.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000056 00000057 00000058 00000059 00000060 00000061 00000062 00000063 00000064 00000065 00000066 00000067 00000068 00000069 00000070 00000071 00000072 00000073 00000074 00000075 00000076 00000077 00000078 00000079 00000080 00000081 00000082 00000083 00000084 00000085 #region constructor /********************\ |* CONSTRUCTOR *| \********************/ public VerifyData (int Algorithm. will be changed to //false if any data verification errors occur verified = true. string GAPopSize. string CCReports. string CCLearningRate. string BPReports.XOR && _TDataSet != Page 111 of 225 .

FISHER && _TDataSet != (int)Dataset.Show ("Parsing Error: Please check supplied cascade" + "correlation data values". } } } Page 112 of 225 . "Parsing Error".William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000086 00000087 00000088 00000089 00000090 00000091 00000092 00000093 00000094 00000095 00000096 00000097 00000098 00000099 00000100 00000101 00000102 00000103 00000104 00000105 00000106 00000107 00000108 00000109 00000110 00000111 00000112 00000113 00000114 00000115 (int)Dataset. verified = false.ToSingle(CCTerr).ToSingle(CCLearningRate).OK.ToInt32(CCReports). if (verified) { if (trainingalg == 2) { try { //Cascade correlation _CC_Terr = Convert. _CC_Reports = Convert. MessageBoxIcon. } catch (FormatException) { MessageBox. _CC_ITerr = CCITerr.CUSTOM) verified = false.Error).ToInt32(CCMaxNeurons). _CC_LearningRate = Convert. _CC_MaxNeurons = Convert. MessageBoxButtons.

ToInt32(GACrossOver). _GA_Crossover = Convert. } } } if (verified) Page 113 of 225 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000116 00000117 00000118 00000119 00000120 00000121 00000122 00000123 00000124 00000125 00000126 00000127 00000128 00000129 00000130 00000131 00000132 00000133 00000134 00000135 00000136 00000137 00000138 00000139 00000140 00000141 00000142 00000143 00000144 00000145 if (verified) { if (trainingalg == 0) { try { //Genetic algorithm _GA_PopSize = Convert. MessageBoxButtons. } catch (FormatException) { MessageBox.Error). _GA_GenLimit = Convert.Show ("Parsing Error: Please check supplied" +" genetic algorithm data values".ToInt32(GAMutation). _GA_ITerr = GAITerr.ToInt32(GAGenLimit).ToSingle(GATerr).OK. "Parsing Error".ToInt32(GAHiddenL). _GA_HiddenL = Convert. MessageBoxIcon. verified = false.ToInt32(GAPopSize). _GA_Reports = Convert.ToInt32(GAReports). _GA_Mutation = Convert. _GA_Terr = Convert.

ToInt32(BPEpochLimit). _BP_HiddenL = Convert. "Parsing Error".ToDouble(BPTerr).Error). verified = false. _BP_ITerr = BPITerr. } catch (FormatException) { MessageBox.OK.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000146 00000147 00000148 00000149 00000150 00000151 00000152 00000153 00000154 00000155 00000156 00000157 00000158 00000159 00000160 00000161 00000162 00000163 00000164 00000165 00000166 00000167 00000168 00000169 00000170 00000171 00000172 00000173 00000174 00000175 { if (trainingalg == 1) { try { //Back-propagation _BP_EpochLimit = Convert.Show ("Parsing Error: Please check " + "supplied back propagation data values". } Page 114 of 225 . _BP_Reports = Convert. } } } if (verified) { verified = checkforsanity(). MessageBoxButtons.ToInt32(BPReports). _BP_Terr = Convert. MessageBoxIcon.ToInt32(BPHiddenL).ToDouble(BPLearningRate). _BP_LearningRate = Convert.

William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000176 00000177 00000178 00000179 00000180 00000181 00000182 00000183 00000184 00000185 00000186 00000187 00000188 00000189 00000190 00000191 00000192 00000193 00000194 00000195 00000196 00000197 00000198 00000199 00000200 00000201 00000202 00000203 00000204 00000205 } #endregion #region internal /********************\ |* INTERNAL *| \********************/ private bool checkGAsanity() { bool sanity = true. if (_GA_GenLimit < 1) sanity = false. } if (_GA_Mutation < 0 || _GA_Mutation > 100) { sanity = false. Page 115 of 225 . if (_GA_PopSize < 3) sanity = false. if (_GA_Crossover < 0 || _GA_Crossover > 100) { sanity = false. } if (_GA_Terr < 0 || _GA_Terr > 1) sanity = false.

Show ("Sanity Check Fail: Please check supplied " + "genetic algorithm data values".OK. if (_BP_EpochLimit < 1) sanity = false. } private bool checkBPsanity() { bool sanity = true.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000206 00000207 00000208 00000209 00000210 00000211 00000212 00000213 00000214 00000215 00000216 00000217 00000218 00000219 00000220 00000221 00000222 00000223 00000224 00000225 00000226 00000227 00000228 00000229 00000230 00000231 00000232 00000233 00000234 00000235 if (_GA_Reports < 1 || _GA_Reports > _GA_GenLimit) sanity = false.Error). MessageBoxButtons. if (_BP_HiddenL < 1) sanity = false. if (_GA_HiddenL < 1) sanity = false. if (_BP_Reports < 1 || _BP_Reports > _BP_EpochLimit) sanity = false. MessageBoxIcon. return sanity. "Sanity Check Fail". if (!sanity) MessageBox. Page 116 of 225 .

OK.Show ("Sanity Check Fail: Please check supplied " + "back propagation data values". MessageBoxIcon. if (_CC_Terr < 0 || _CC_Terr > 1) sanity = false.Error). MessageBoxButtons.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000236 00000237 00000238 00000239 00000240 00000241 00000242 00000243 00000244 00000245 00000246 00000247 00000248 00000249 00000250 00000251 00000252 00000253 00000254 00000255 00000256 00000257 00000258 00000259 00000260 00000261 00000262 00000263 00000264 00000265 if (_BP_Terr < 0 || _BP_Terr > 1) sanity = false. "Sanity Check Fail". return sanity. if (_BP_LearningRate < 0 || _BP_LearningRate > 1) sanity = false. if (_CC_Reports < 1 || _CC_Reports > _CC_MaxNeurons) sanity = false. } private bool checkCCsanity() { bool sanity = true. if (_CC_MaxNeurons < 1) sanity = false. Page 117 of 225 . if (!sanity) MessageBox.

only //halt any values from proceeding that are so wrong they could cause //serious errors later down the line. We're also doing some very minor processing of the probability values. return sanity. "Sanity Check Fail". if (trainingalg < 0 || trainingalg > 2) Page 118 of 225 . } private bool checkforsanity() { //Some of the values that are allowed through here would not produce //good results . MessageBoxIcon.OK. MessageBoxButtons.Error).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000266 00000267 00000268 00000269 00000270 00000271 00000272 00000273 00000274 00000275 00000276 00000277 00000278 00000279 00000280 00000281 00000282 00000283 00000284 00000285 00000286 00000287 00000288 00000289 00000290 00000291 00000292 00000293 00000294 00000295 if (_CC_LearningRate < 0 || _CC_LearningRate > 1) sanity = false. if they're greater than 1 but less than 100 scale them appropriately.however I don't want to prevent any flexibility. // // // // // bool Basically we're making sure the data is clean before it goes anywhere near a DLL.Show ("Sanity Check Fail: Please check supplied " + "cascade correlation data values". sanity = true. if (!sanity) MessageBox.

} } return sanity.Show ("Sanity Check Fail: Training Algorithm not " + "selected or selected incorrectly". } if (trainingalg == 1) { sanity = checkBPsanity(). } Page 119 of 225 . MessageBoxIcon. } if (trainingalg == 2) { sanity = checkCCsanity(). MessageBoxButtons.OK.Error). } if (sanity) { if (trainingalg == 0) { sanity = checkGAsanity().William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000296 00000297 00000298 00000299 00000300 00000301 00000302 00000303 00000304 00000305 00000306 00000307 00000308 00000309 00000310 00000311 00000312 00000313 00000314 00000315 00000316 00000317 00000318 00000319 00000320 00000321 00000322 00000323 00000324 00000325 { sanity = false. "Sanity Check Fail". MessageBox.

} } public int GA_Crossover { get { return _GA_Crossover. } } #endregion #region genetic_algorithm //Genetic Algorithm public int GA_PopSize { get { return _GA_PopSize. } } public int GA_GenLimit { get { return _GA_GenLimit. } } public int CC_Reports { get { return _CC_Reports. } } public bool CC_ITerr { get { return _CC_ITerr. } } public int CC_MaxNeurons { get { return _CC_MaxNeurons.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000326 00000327 00000328 00000329 00000330 00000331 00000332 00000333 00000334 00000335 00000336 00000337 00000338 00000339 00000340 00000341 00000342 00000343 00000344 00000345 00000346 00000347 00000348 00000349 00000350 00000351 00000352 00000353 00000354 00000355 #endregion #region getters/setters /********************\ |* EXTERNAL *| \********************/ #region cascade_correlation //Cascade Correlation public double CC_Terr { get { return _CC_Terr. } } public double CC_LearningRate { get { return _CC_LearningRate. } } public int GA_Mutation Page 120 of 225 .

} } _GA_HiddenL. } } public int BP_Reports { get { return _BP_Reports. } } _GA_Reports. } } public bool BP_ITerr { get { return _BP_ITerr. } } public int BP_HiddenL { get { return _BP_HiddenL. } } _GA_Terr. } } public double BP_Terr { get { return _BP_Terr.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000356 00000357 00000358 00000359 00000360 00000361 00000362 00000363 00000364 00000365 00000366 00000367 00000368 00000369 00000370 00000371 00000372 00000373 00000374 00000375 00000376 00000377 00000378 00000379 00000380 00000381 00000382 00000383 00000384 00000385 { get { return public double GA_Terr { get { return public int GA_Reports { get { return public int GA_HiddenL { get { return public bool GA_ITerr { get { return #endregion _GA_Mutation. } } _GA_ITerr. } } #endregion #endregion #region external //Verification Page 121 of 225 . } } public double BP_LearningRate { get { return _BP_LearningRate. } } #region back_propagation //Back propagation public int BP_EpochLimit { get { return _BP_EpochLimit.

William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000386 00000387 00000388 00000389 00000390 00000391 00000392 00000393 00000394 00000395 00000396 00000397 00000398 } public bool Verified { get { return verified. } } #endregion } Page 122 of 225 . } } //TrainingAlgorithm public int TrainingAlgorithm { get { return trainingalg. } } //Dataset public int TrDataSet { get { return _TDataSet.

private ArrayList teinputs. Page 123 of 225 .2. private ArrayList trinputs.Generic.Collections. namespace _05025397.Collections.Forms. private string[] data.Text. System.cs” 00000001 00000002 00000003 00000004 00000005 00000006 00000007 00000008 00000009 00000010 00000011 00000012 00000013 00000014 00000015 00000016 00000017 00000018 00000019 00000020 00000021 00000022 00000023 00000024 00000025 00000026 00000027 00000028 00000029 using using using using using using System. System.Windows. System.1.Linq. private ArrayList aldata. private ArrayList altest.Controller { public class DatasetParser { #region data /********************\ |* DATA *| \********************/ private bool _Verified. System.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 9.“DatasetParser. private ArrayList troutputs. private ArrayList teoutputs. private string[] test.2 . System.

RichTextBox txtTest) { //Data init bool Verified = true.Text = txtTest. teoutputs = new ArrayList(). troutputs = new ArrayList().Text = txtData. private double Noutputs. txtTest.Text.Replace("\r". '\t'.'.Text = txtTest.' }. aldata = new ArrayList(). ""). ' '. '. #endregion #region constructor /********************\ |* CONSTRUCTOR *| \********************/ public DatasetParser(RichTextBox txtData. txtData. Page 124 of 225 .Trim(). '.Trim(). teinputs = new ArrayList().Text = txtData.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000030 00000031 00000032 00000033 00000034 00000035 00000036 00000037 00000038 00000039 00000040 00000041 00000042 00000043 00000044 00000045 00000046 00000047 00000048 00000049 00000050 00000051 00000052 00000053 00000054 00000055 00000056 00000057 00000058 00000059 //private Scaler scale. txtTest.Replace("\r".Text. trinputs = new ArrayList(). '\r'. private double Ninputs. char[] delimiterChars = { '\n'.Text. altest = new ArrayList(). //Strip carriage returns and trim txtData. ""). private double Niterations.Text.

try { for (int i = 3. Noutputs = System. } #endregion #region internal_functions /********************\ |* INTERNAL *| \********************/ private bool parseTrainingData() { bool success = true. number of inputs //and number of outputs Niterations = System.Length. i++) Page 125 of 225 . Ninputs = System. _Verified = Verified.Text.ToDouble(data[1]).Split(delimiterChars).ToDouble(data[2]). //parse the data into appropriately //structured arrays Verified = parseTrainingData().Split(delimiterChars).Convert.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000060 00000061 00000062 00000063 00000064 00000065 00000066 00000067 00000068 00000069 00000070 00000071 00000072 00000073 00000074 00000075 00000076 00000077 00000078 00000079 00000080 00000081 00000082 00000083 00000084 00000085 00000086 00000087 00000088 00000089 //Split at the chars specified above data = txtData. //Get the number of iterations.ToDouble(data[0]).Convert.Convert. i < data.Text. Verified = parseTestingData(). test = txtTest.

RemoveAt(0). success = false. } } } catch { MessageBox. aldata. } return success. MessageBoxIcon.Add(aldata[0]).Add(System.Show("Error parsing training data".Error). j < Noutputs. } for (int j = 0. i < Ninputs. aldata. j++) { troutputs. } private bool parseTestingData() { bool success = true.OK.Convert. Page 126 of 225 . "Error".ToDouble(data[i])).Add(aldata[0]). i++) { trinputs.Count > 0) { for (int i = 0.RemoveAt(0). } while (aldata. MessageBoxButtons.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000090 00000091 00000092 00000093 00000094 00000095 00000096 00000097 00000098 00000099 00000100 00000101 00000102 00000103 00000104 00000105 00000106 00000107 00000108 00000109 00000110 00000111 00000112 00000113 00000114 00000115 00000116 00000117 00000118 00000119 { aldata.

"Error".Add(System.Add(altest[0]). j++) { teoutputs. MessageBoxButtons. i++) { altest.RemoveAt(0).Add(altest[0]). i < Ninputs. } return success.RemoveAt(0).ToDouble(data[i])).Error).Length. MessageBoxIcon. } for (int j = 0.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000120 00000121 00000122 00000123 00000124 00000125 00000126 00000127 00000128 00000129 00000130 00000131 00000132 00000133 00000134 00000135 00000136 00000137 00000138 00000139 00000140 00000141 00000142 00000143 00000144 00000145 00000146 00000147 00000148 00000149 try { for (int i = 3.Count > 0) { for (int i = 0. i++) { teinputs. } Page 127 of 225 . success = false. j < Noutputs. altest. altest. } while (altest.Convert.Show("Error parsing testing data". } } } catch { MessageBox.OK. i < data.

troutputdata = new double[troutputs. i < (troutputs. i < (trinputs. out double[][] troutputdata) { trinputdata = new double[trinputs. for (int j = 0. i++) troutputdata[i] = new double[(int) Noutputs]. k++) { Page 128 of 225 .Count / (int)Noutputs).Count / (int)Ninputs). j < trinputs.Count / (int) Ninputs][]. j+=(int) Ninputs.Count / (int) Noutputs][]. for (int i = 0.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000150 00000151 00000152 00000153 00000154 00000155 00000156 00000157 00000158 00000159 00000160 00000161 00000162 00000163 00000164 00000165 00000166 00000167 00000168 00000169 00000170 00000171 00000172 00000173 00000174 00000175 00000176 00000177 00000178 00000179 #endregion #region external_functions /********************\ |* EXTERNAL *| \********************/ public void GetStructure(out double iter. for (int i = 0. out double Ninp. l = 0. i++) trinputdata[i] = new double[(int) Ninputs]. } public void GetTrainingData (out double[][] trinputdata.Count. iter = Niterations. out double Nout) { Ninp = Ninputs. Nout = Noutputs. l++) { for (int k = 0. k < Ninputs.

for (int j = 0.Count / (int)Ninputs). j += (int)Ninputs. k < Ninputs. k]. k++) { teinputdata[l][k] = (double)teinputs[j + k]. i++) teoutputdata[i] = new double[(int)Noutputs]. teoutputdata = new double[teoutputs. out double[][] teoutputdata) { teinputdata = new double[teinputs. k < Noutputs. l = 0. j < teinputs. j < troutputs. l = 0. } } } public void GetTestingData (out double[][] teinputdata. l++) { for (int k = 0.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000180 00000181 00000182 00000183 00000184 00000185 00000186 00000187 00000188 00000189 00000190 00000191 00000192 00000193 00000194 00000195 00000196 00000197 00000198 00000199 00000200 00000201 00000202 00000203 00000204 00000205 00000206 00000207 00000208 00000209 trinputdata[l][k] = (double) trinputs[j + } } for (int j = 0. i < (teoutputs. for (int i = 0.Count / (int)Noutputs).Count / (int)Ninputs][]. i < (teinputs.Count. i++) teinputdata[i] = new double[(int)Ninputs]. Page 129 of 225 . j += (int)Noutputs. for (int i = 0. l++) { for (int k = 0.Count.Count / (int)Noutputs][]. k++) { troutputdata[l][k] = (double)troutputs[j + k].

3 . k++) { teoutputdata[l][k] = (double)teoutputs[j + k].“Algorithms\TrainingAlgorithm.1.2. k < Noutputs.Count. j < teoutputs. l = 0. Page 130 of 225 . } } #endregion } 9.cs” 00000001 using System. } } } public void GetTestingDataAL (out ArrayList teinputdata. } public bool Verified { get { return _Verified. j += (int)Noutputs. l++) { for (int k = 0. out ArrayList teoutputdata) { teinputdata = teinputs.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000210 00000211 00000212 00000213 00000214 00000215 00000216 00000217 00000218 00000219 00000220 00000221 00000222 00000223 00000224 00000225 00000226 00000227 00000228 00000229 00000230 00000231 00000232 00000233 00000234 00000235 00000236 } } } for (int j = 0. teoutputdata = teoutputs.

Text.Algorithms { public abstract class TrainingAlgorithm { #region data /********************\ |* DATA *| \********************/ private double _TargetError. private int _ReportInterval. ZedGraph.Windows. protected bool _Trained. private string _ReportData. _05025397.Linq.Collections. namespace _05025397. Page 131 of 225 . protected bool _Tested. private int _IterMax. protected PointPairList BestErrList. protected int CurIt.Controller. System. protected bool _ITerr. System. System.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000002 00000003 00000004 00000005 00000006 00000007 00000008 00000009 00000010 00000011 00000012 00000013 00000014 00000015 00000016 00000017 00000018 00000019 00000020 00000021 00000022 00000023 00000024 00000025 00000026 00000027 00000028 00000029 00000030 00000031 using using using using using using System.Forms.Generic.

RichTextBox txtData. #endregion #region constructor /********************\ |* CONSTRUCTOR *| \********************/ public TrainingAlgorithm(double Target_Error. so the //network will loop to completion. bool ITerr) { _TargetError = Target_Error. if (_ITerr) _TargetError = -1. //If the ignore target error is true //then set the target error to -1. int Iter_Max. _IterMax = Iter_Max. int Reports.00. _ITerr = ITerr. getdata = new DatasetParser(txtData. _ReportInterval = Reports. txtTest). RichTextBox txtTest. BestErrList = new PointPairList(). } #endregion Page 132 of 225 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000032 00000033 00000034 00000035 00000036 00000037 00000038 00000039 00000040 00000041 00000042 00000043 00000044 00000045 00000046 00000047 00000048 00000049 00000050 00000051 00000052 00000053 00000054 00000055 00000056 00000057 00000058 00000059 00000060 00000061 protected Random Rand = new Random().0 //this is impossible to reach. protected DatasetParser getdata.

} get { return _TargetError. } #endregion #region external /********************\ |* EXTERNAL *| \********************/ public abstract bool Train(). } } Page 133 of 225 . public abstract bool Test(). public abstract string network_details().William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000062 00000063 00000064 00000065 00000066 00000067 00000068 00000069 00000070 00000071 00000072 00000073 00000074 00000075 00000076 00000077 00000078 00000079 00000080 00000081 00000082 00000083 00000084 00000085 00000086 00000087 00000088 00000089 00000090 00000091 #region internal /********************\ |* INTERNAL *| \********************/ protected void Report(string output) { _ReportData += output. } public double TargetError { set { _TargetError = value. public PointPairList getGraphData() { return BestErrList.

} } public int IterMax { get { return _IterMax. } } public bool Trained { get { return _Trained. } set { _Trained = value.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000092 00000093 00000094 00000095 00000096 00000097 00000098 00000099 00000100 00000101 00000102 00000103 00000104 00000105 00000106 00000107 00000108 00000109 00000110 00000111 00000112 00000113 00000114 00000115 00000116 00000117 00000118 00000119 } public int ReportInterval { get { return _ReportInterval. } } #endregion } Page 134 of 225 . } } public string ReportData { get { return _ReportData. } } public bool Tested { get { return _Tested.

Linq. private double _LearnRate.Algorithms { public class CascadeCorrelation : TrainingAlgorithm { #region data /********************\ |* DATA *| \********************/ private double _InputL. FANN_Cascade CCNet.Collections. System.Windows. private double _OutputL. System.cs” 00000001 00000002 00000003 00000004 00000005 00000006 00000007 00000008 00000009 00000010 00000011 00000012 00000013 00000014 00000015 00000016 00000017 00000018 00000019 00000020 00000021 00000022 00000023 00000024 00000025 00000026 00000027 00000028 00000029 using using using using using using using using using using System. System. _05025397.IO.Generic.2. System. namespace _05025397. #endregion #region constructor Page 135 of 225 .Text. System. FANN_Wrapper.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 9.Controller.1.Collections.Forms.4 .“Algorithms\CascadeCorrelation. ZedGraph. System.

txtTest.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000030 00000031 00000032 00000033 00000034 00000035 00000036 00000037 00000038 00000039 00000040 00000041 00000042 00000043 00000044 00000045 00000046 00000047 00000048 00000049 00000050 00000051 00000052 00000053 00000054 00000055 00000056 00000057 00000058 00000059 /********************\ |* CONSTRUCTOR *| \********************/ public CascadeCorrelation(double LearnRate. Reports. _LearnRate. out _OutputL). double TargetError. RichTextBox te) { Page 136 of 225 . (int)_OutputL. base.ReportInterval. base.GetStructure(out iterations. bool ITerr. double iterations. ITerr) { _LearnRate = LearnRate. int Iter_Max. int Reports. saveDllData(txtData.TargetError. RichTextBox txtData. out _InputL. CCNet = new FANN_Cascade((int)_InputL. base.IterMax). RichTextBox txtTest) : base(TargetError. getdata. Iter_Max. txtData. } #endregion #region internal_functions /********************\ |* INTERNAL *| \********************/ private void saveDllData(RichTextBox tr. txtTest).

Close(). tw.dat").William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000060 00000061 00000062 00000063 00000064 00000065 00000066 00000067 00000068 00000069 00000070 00000071 00000072 00000073 00000074 00000075 00000076 00000077 00000078 00000079 00000080 00000081 00000082 00000083 00000084 00000085 00000086 00000087 00000088 00000089 TextWriter tw = new StreamWriter("dlltrdata.00000}".Append( string.Close().Write(tr. tw. output. tw. (int)_InputL)).Format ("\nLearning Rate: {0:0.Format ("Network Details: \n Input Layer: {0:d}".Text). _LearnRate)).Append( string.dat"). (int)_OutputL)).Write(te.Text).Append( string.Append( Page 137 of 225 . output. output. } #endregion #region external_functions /********************\ |* EXTERNAL *| \********************/ public override string network_details() { StringBuilder output = new StringBuilder().Format ("\t\t\t\t\tOutput Layer: {0:d}". output. tw = new StreamWriter("dlltedata. tw.

} public override bool Train() { bool success = true. Report(CCNet.Format ("\nReport every {0:d} generations. output.".William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000090 00000091 00000092 00000093 00000094 00000095 00000096 00000097 00000098 00000099 00000100 00000101 00000102 00000103 00000104 00000105 00000106 00000107 00000108 00000109 00000110 00000111 00000112 00000113 00000114 00000115 00000116 00000117 00000118 00000119 string._tr_output).Format ("\t\t\t\t\t\tIgnore target error " + "(process all iterations): {0}". try { success = CCNet.ReportInterval)). output. return output. base. } catch { Page 138 of 225 . output. base. _ITerr)).TargetError)).ToString().Append( string.00000}".Format ("\t\t\tTarget Error: {0:0.Format ("\t\tMaximum Nodes: {0:d}".IterMax)).Append( string.Train(). base.Append( string.

success = false. "Error". besterrlist[i]).Show("Error getting graph data from DLL". MessageBoxButtons.BestErrList. } Page 139 of 225 . base.OK.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000120 00000121 00000122 00000123 00000124 00000125 00000126 00000127 00000128 00000129 00000130 00000131 00000132 00000133 00000134 00000135 00000136 00000137 00000138 00000139 00000140 00000141 00000142 00000143 00000144 00000145 00000146 00000147 00000148 00000149 MessageBox. "Error". } } catch { MessageBox. MessageBoxButtons. MessageBoxIcon._Trained = success.besterrorlist. MessageBoxIcon.OK. i < besterrlist.Show("Error running Cascade training!". success = false. } public bool getGraphDataFromDll() { bool success = true.Error). i++) { base. } success = getGraphDataFromDll(). try { double[] besterrlist = CCNet. return success.Length.Error). for (int i = 0.Add((double)i.

OK. try { success = CCNet._te_output). Report(CCNet.Error). success = false. MessageBoxButtons. } #endregion } Page 140 of 225 . return success._Tested = success. } catch { MessageBox. } public override bool Test() { bool success = true.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000150 00000151 00000152 00000153 00000154 00000155 00000156 00000157 00000158 00000159 00000160 00000161 00000162 00000163 00000164 00000165 00000166 00000167 00000168 00000169 00000170 00000171 00000172 00000173 } return success.Test(). } base. "Error". MessageBoxIcon.Show("Error running Cascade testing!".

private double _OutputL.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 9.cs” 00000001 00000002 00000003 00000004 00000005 00000006 00000007 00000008 00000009 00000010 00000011 00000012 00000013 00000014 00000015 00000016 00000017 00000018 00000019 00000020 00000021 00000022 00000023 00000024 00000025 00000026 00000027 00000028 00000029 using using using using using using using using System. //Network Data private double _InputL. _05025397.Collections. System. private int _Crossover.“Algorithms\GeneticAlgorithm.2. System.Algorithms { public class GeneticAlgorithm : TrainingAlgorithm { #region data /********************\ |* DATA *| \********************/ //GA Data private int _PopSize. System. System. namespace _05025397.Windows. ZedGraph. System. //Population Page 141 of 225 .Linq.Controller.Forms. private int _Mutation.Text.Collections.1.Generic. private int _HiddenL.5 .

txtData. int HiddenL. Iter_Max. i++) { Population. getdata. ITerr) { _PopSize = PopSize. out _OutputL). Reports. RichTextBox txtData. //Init Network params double iterations. int Reports. txtTest. _Mutation = Mutation. //Create new population references Population = new ArrayList(_PopSize). #endregion #region constructor /********************\ |* CONSTRUCTOR *| \********************/ public GeneticAlgorithm(int PopSize.Add Page 142 of 225 . out _InputL. for (int i = 0. RichTextBox txtTest) : base(TargetError. int Mutation. bool ITerr. int Crossover.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000030 00000031 00000032 00000033 00000034 00000035 00000036 00000037 00000038 00000039 00000040 00000041 00000042 00000043 00000044 00000045 00000046 00000047 00000048 00000049 00000050 00000051 00000052 00000053 00000054 00000055 00000056 00000057 00000058 00000059 ArrayList Population. _HiddenL = HiddenL. i < _PopSize. int Iter_Max. _Crossover = Crossover.GetStructure(out iterations. double TargetError.

Page 143 of 225 .GetScaledTrainingData(out inputs. getdata.IterMax. Rand. for (int i = 0. j < _PopSize.Next())). _HiddenL.getMsQE(inputs. i++) { //Get the msqe for (int j = 0. } } #endregion #region internal_functions /********************\ |* INTERNAL *| \********************/ private bool run() { bool success = true. double[][] results.GANetwork) Population[j]). double[][] inputs. } //Sort the functions according to fitness now. out results). success = sort_fitnesses(). (int)_OutputL.GANetwork( (int)_InputL. i < base.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000060 00000061 00000062 00000063 00000064 00000065 00000066 00000067 00000068 00000069 00000070 00000071 00000072 00000073 00000074 00000075 00000076 00000077 00000078 00000079 00000080 00000081 00000082 00000083 00000084 00000085 00000086 00000087 00000088 00000089 (new Model. j++) { ((Model. results).

\n").MSqE).GANetwork)Population[0]).GANetwork)Population[0]). //Check for having reached the target if (((Model.Report("Best error at iteration [ " + i + " ] was " + ((Model. ((Model.CurIt == ReportInterval) { base. } base. } private bool sort_fitnesses() { Page 144 of 225 .").Add ((double)i. break.CurIt = 0.TargetError) { Report("\nNetwork matching or improving upon target error"+ "found at iteration [ " + i + " ]. base.BestErrList. } success = build_generation().GANetwork)Population[0]).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000090 00000091 00000092 00000093 00000094 00000095 00000096 00000097 00000098 00000099 00000100 00000101 00000102 00000103 00000104 00000105 00000106 00000107 00000108 00000109 00000110 00000111 00000112 00000113 00000114 00000115 00000116 00000117 00000118 00000119 //Report if we're at a report iteration //Also update the besterrlist (for graphing) if (base. } return success.CurIt++.MSqE + ". base.MSqE <= base.

William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000120 00000121 00000122 00000123 00000124 00000125 00000126 00000127 00000128 00000129 00000130 00000131 00000132 00000133 00000134 00000135 00000136 00000137 00000138 00000139 00000140 00000141 00000142 00000143 00000144 00000145 00000146 00000147 00000148 00000149 bool success = true.Error). double[] weightarray2. MessageBoxButtons.Show("Error sorting population". double[] weightarray1. i > (int)(_PopSize / 2). } catch { MessageBox. selection2. success = false. for (int i = (_PopSize . "Error".1). } return success. try { Population. //It selects two points from the start of the population //Then applies crossover if necessary (based on probability) Page 145 of 225 . } private bool build_generation() { bool success = true. int selection1.OK. double[] weightarray_f. i--) { //This algorithm is altered from previous versions. MessageBoxIcon.Sort().

Next((int)_PopSize / 2). //Next iteration it replaces the next to last.Length]. //Then it replaces the last position in the population //(ie. success = false. selection2 = Rand.Length) { MessageBox. weightarray2 = ((Model.". if (Rand.getWeights().getWeights(). selection1 = Rand. "Error".Length != weightarray2. the least fit) with this new string. MessageBoxButtons.GANetwork)Population[selection1]).Error). and //so on.OK.Next((int)_PopSize / 2). MessageBoxIcon.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000150 00000151 00000152 00000153 00000154 00000155 00000156 00000157 00000158 00000159 00000160 00000161 00000162 00000163 00000164 00000165 00000166 00000167 00000168 00000169 00000170 00000171 00000172 00000173 00000174 00000175 00000176 00000177 00000178 00000179 //applies mutation if necessary (same again) //then if neither mutation or Xover have been applied //chooses position 1 or 2.GANetwork)Population[selection1]). all the way to the middle of the population. weightarray1 = ((Model. //Just a quick error check if (weightarray1.Next(100) < _Crossover) { //Choose a random position in the weight array //to be our crossover point Page 146 of 225 . } try { weightarray_f = new double[weightarray1.Show("Error: 2D weight array length unequal.

} else { weightarray_f[j] = weightarray2[j]. } Page 147 of 225 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000180 00000181 00000182 00000183 00000184 00000185 00000186 00000187 00000188 00000189 00000190 00000191 00000192 00000193 00000194 00000195 00000196 00000197 00000198 00000199 00000200 00000201 00000202 00000203 00000204 00000205 00000206 00000207 00000208 00000209 int XPos = Rand. (weightarray1. //Simple single point crossover of the weights for (int j = 0.Next(100) < _Mutation) { //Jiggle the weights in the array //up or down by 0. } } //Set the weights of the current member of the //population we're on ((Model. } } } if (Rand.Length .1)).5.Length. j++) { weightarray_f[j] += Rand.GANetwork)Population[i]).NextDouble() .Length.5 for (int j = 0. j++) { if (j < XPos) { weightarray_f[j] = weightarray1[j].0.Next(1.setWeights(weightarray_f). j < weightarray1. j < weightarray_f.

"Error". } } return success.OK.Append( Page 148 of 225 . success = false.Show("Error building new generation". _Mutation)). output. output. output. } #endregion #region external_functions /********************\ |* EXTERNAL *| \********************/ public override string network_details() { //Add stringbuilder for efficiency here StringBuilder output = new StringBuilder().Append( string. output.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000210 00000211 00000212 00000213 00000214 00000215 00000216 00000217 00000218 00000219 00000220 00000221 00000222 00000223 00000224 00000225 00000226 00000227 00000228 00000229 00000230 00000231 00000232 00000233 00000234 00000235 00000236 00000237 00000238 00000239 catch { MessageBox.Error). MessageBoxButtons. MessageBoxIcon.Format("\t\t\tCrossover Probability: {0:d}". _PopSize)). _Crossover)).Format("Network Details: \n Population: {0:d}".Append( string.Format("\t\t\tMutation Probability: {0:d}".Append( string.

} public override bool Train() { bool success = true.Format("\t\t\tTarget Error: {0:0.IterMax)).Format("\nReport every {0:d} generations. } public override bool Test() Page 149 of 225 . output. base.00000}".Append( string. output.ReportInterval)). return success.Format("\t\t\t\t\t\t\tIgnore target error (process all iterations): {0}".Append( string.Format("\t\t\tMaximum Generation: {0:d}". _HiddenL)).TargetError)).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000240 00000241 00000242 00000243 00000244 00000245 00000246 00000247 00000248 00000249 00000250 00000251 00000252 00000253 00000254 00000255 00000256 00000257 00000258 00000259 00000260 00000261 00000262 00000263 00000264 00000265 00000266 00000267 00000268 00000269 string.Append( string. output. output. base. base.Append( string. base.ToString().Format("\nHidden layer: {0:d}". success = run()._Trained = success. return output. _ITerr)).".

out tstoutput).Format(" [{0:g}] ".Length. tstinput[i][j])).Format(" [{0:g}] ". i++) { ((Model.GANetwork)Population[0]).Append(string. output.Append("\nWith inputs:"). j++) { output. Page 150 of 225 . j++) { output. output. double[] netoutput.GANetwork)Population[0]). } output. for (int j = 0.Length.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000270 00000271 00000272 00000273 00000274 00000275 00000276 00000277 00000278 00000279 00000280 00000281 00000282 00000283 00000284 00000285 00000286 00000287 00000288 00000289 00000290 00000291 00000292 00000293 00000294 00000295 00000296 00000297 00000298 00000299 { bool success = true.Append(string. j < netoutput. getdata. double[][] tstinput.GetScaledTestingData(out tstinput. double[][] tstoutput. netoutput = ((Model. j < tstinput[i].GetOutput().Append("\n\n\t\t\t~~Network Testing~~"). try { for (int i = 0. i < tstinput.Append("\nOutput achieved was:"). StringBuilder output = new StringBuilder(). for (int j = 0.Length.Run(tstinput[i]). netoutput[j])).

Format ("{0:g}". j++) { output. base. for (int j = 0. } catch { MessageBox.getMsQE(tstinput. (double) ((Model. } #endregion } Page 151 of 225 .Show("Error Running Test. output.Error).OK.GANetwork)Population[0]).Append(string. tstoutput[i][j])).Append("\nMean Squared Error with these inputs and outputs is:").tstoutput). MessageBoxButtons. j < tstoutput[i].Append( string. success = false.Length.Report(output.MSqE)). } } output.ToString()). "Error". MessageBoxIcon. ((Model.Format(" [{0:g}] "._Tested = success. } base.". return success.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000300 00000301 00000302 00000303 00000304 00000305 00000306 00000307 00000308 00000309 00000310 00000311 00000312 00000313 00000314 00000315 00000316 00000317 00000318 00000319 00000320 00000321 00000322 00000323 00000324 00000325 00000326 00000327 00000328 } } output.GANetwork)Population[0]).Append("\nOutput Desired was:").

private int _HiddenL.IO.Controller.cs” 00000001 00000002 00000003 00000004 00000005 00000006 00000007 00000008 00000009 00000010 00000011 00000012 00000013 00000014 00000015 00000016 00000017 00000018 00000019 00000020 00000021 00000022 00000023 00000024 00000025 00000026 00000027 00000028 00000029 using using using using using using using using using using System. System. System. private double _OutputL. System.Linq. FANN_Wrapper. System.Text.Generic. _05025397.“Algorithms\BackProp. ZedGraph.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 9. System. #endregion #region constructor Page 152 of 225 .Windows.Algorithms { public class BackProp : TrainingAlgorithm { #region data /********************\ |* DATA *| \********************/ private double _InputL.Collections.Forms.6 .2. System. namespace _05025397. private double _LearnRate.1.Collections. FANN_BackProp BPnet.

out _InputL. } #endregion #region internal_functions /********************\ Page 153 of 225 . (int)_OutputL. out _OutputL). RichTextBox txtData.IterMax). int Reports. txtData. base. base. txtTest. txtTest). double TargetError. Iter_Max. _LearnRate = LearnRate.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000030 00000031 00000032 00000033 00000034 00000035 00000036 00000037 00000038 00000039 00000040 00000041 00000042 00000043 00000044 00000045 00000046 00000047 00000048 00000049 00000050 00000051 00000052 00000053 00000054 00000055 00000056 00000057 00000058 00000059 /********************\ |* CONSTRUCTOR *| \********************/ public BackProp (int HiddenL.TargetError. ITerr) { _HiddenL = HiddenL.ReportInterval. base. saveDllData(txtData. Reports. //Copy the training and testing data to the dat files //ready for the DLL's to access. bool ITerr. _LearnRate. BPnet = new FANN_BackProp((int)_InputL. double LearnRate. RichTextBox txtTest) : base(TargetError.GetStructure(out iterations. double iterations. getdata. HiddenL. int Iter_Max.

William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000060 00000061 00000062 00000063 00000064 00000065 00000066 00000067 00000068 00000069 00000070 00000071 00000072 00000073 00000074 00000075 00000076 00000077 00000078 00000079 00000080 00000081 00000082 00000083 00000084 00000085 00000086 00000087 00000088 00000089 |* INTERNAL *| \********************/ private void saveDllData(RichTextBox tr. output. tw.Write(te. tw. tw = new StreamWriter("dlltedata.dat").Text). tw.Close().Close().Append( string. } #endregion #region external_functions /********************\ |* EXTERNAL *| \********************/ public override string network_details() { StringBuilder output = new StringBuilder(). RichTextBox te) { TextWriter tw = new StreamWriter("dlltrdata.Text).Append( string. tw.Format ("Network Details: \n Input Layer: {0:d}".Write(tr. (int) _OutputL)). output.dat").Append( Page 154 of 225 . output.Format ("\t\t\tOutput Layer: {0:d}". (int) _InputL)).

(int) _HiddenL)). output.00000}".Append( string.Format ("\nLearning Rate: {0:0. _LearnRate)).Format ("\t\t\t\t\t\tIgnore target error " + "(process all iterations): {0}".Append( string. output. } public override bool Train() { Page 155 of 225 .Format ("\t\tMaximum Epoch: {0:d}". output. base. output.Append( string.Format ("\t\t\tTarget Error: {0:0.". base.Format ("\t\t\tHidden Layer: {0:d}".Format ("\nReport every {0:d} generations.ToString().00000}". return output.Append( string. base.IterMax)).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000090 00000091 00000092 00000093 00000094 00000095 00000096 00000097 00000098 00000099 00000100 00000101 00000102 00000103 00000104 00000105 00000106 00000107 00000108 00000109 00000110 00000111 00000112 00000113 00000114 00000115 00000116 00000117 00000118 00000119 string.TargetError)).ReportInterval)). output.Append( string. _ITerr)).

for (int i = 0. besterrlist[i]). return success.Add((double)i._Trained = success. "Error". } success = getGraphDataFromDll().Show("Error running Backprop training!". Report(BPnet. MessageBoxIcon. try { success = BPnet. success = false.Train(). base. } catch { MessageBox.OK. Page 156 of 225 . i < besterrlist.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000120 00000121 00000122 00000123 00000124 00000125 00000126 00000127 00000128 00000129 00000130 00000131 00000132 00000133 00000134 00000135 00000136 00000137 00000138 00000139 00000140 00000141 00000142 00000143 00000144 00000145 00000146 00000147 00000148 00000149 bool success = true.besterrorlist. try { double[] besterrlist = BPnet. i++) { base.Length. MessageBoxButtons. } public bool getGraphDataFromDll() { bool success = true.BestErrList.Error)._tr_output).

Error). "Error". } #endregion Page 157 of 225 . success = false.Error). MessageBoxIcon. success = false._Tested = success. MessageBoxButtons.Show("Error getting graph data from DLL".OK. } return success.Test(). } base.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000150 00000151 00000152 00000153 00000154 00000155 00000156 00000157 00000158 00000159 00000160 00000161 00000162 00000163 00000164 00000165 00000166 00000167 00000168 00000169 00000170 00000171 00000172 00000173 00000174 00000175 00000176 00000177 00000178 00000179 } } catch { MessageBox. return success. MessageBoxIcon.OK. "Error". } public override bool Test() { bool success = true._te_output). MessageBoxButtons.Show("Error running Backprop testing!". } catch { MessageBox. Report(BPnet. try { success = BPnet.

Forms. _05025397.1. Page 158 of 225 . protected int _numinputs.Text.3 .Windows.Generic.“Network.Model { public class Network { #region data /********************\ |* DATA *| \********************/ //[layer][node][input] protected double[][][] _inputs. protected double[][][] _weights.3. System.Linq. protected int _hiddennodes. System.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000180 00000181 } } 9.Model 9. namespace _05025397. System.Collections.1 .cs” 00000001 00000002 00000003 00000004 00000005 00000006 00000007 00000008 00000009 00000010 00000011 00000012 00000013 00000014 00000015 00000016 00000017 00000018 00000019 00000020 00000021 00000022 00000023 00000024 using using using using using using System.1. System. protected double[][] _outputs. protected int _numoutputs.

_weights[1] = new double[_hiddennodes][]. _weights[0] = new double[_numinputs][]. _outputs[1] = new double[_hiddennodes]. Page 159 of 225 . for (int i = 0.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000025 00000026 00000027 00000028 00000029 00000030 00000031 00000032 00000033 00000034 00000035 00000036 00000037 00000038 00000039 00000040 00000041 00000042 00000043 00000044 00000045 00000046 00000047 00000048 00000049 00000050 00000051 00000052 00000053 00000054 protected Random Rand. _outputs[0] = new double[_numinputs]. #endregion #region init /********************\ |* INIT *| \********************/ private void init_inputlayer() { _inputs[0] = new double[_numinputs][]. i < _hiddennodes. for (int i = 0. i++) { _inputs[0][i] = new double[2]. i++) { _inputs[1][i] = new double[_numinputs + 1]. i < _numinputs. _weights[0][i] = new double[2]. } } private void init_hiddenlayer() { _inputs[1] = new double[_hiddennodes][]. _weights[1][i] = new double[_numinputs + 1].

i++) { _inputs[1][i][_numinputs] = 1. i < _numinputs. i < _numoutputs. } //Hidden layer bias for (int i = 0.0. _outputs[2] = new double[_numoutputs]. i++) { _inputs[2][i] = new double[_hiddennodes + 1].0. _weights[2][i] = new double[_hiddennodes + 1]. _weights[2] = new double[_numoutputs][].0.0. _weights[1][i][_numinputs] = 1. } } private void init_bias() { //Input bias for (int i = 0. Page 160 of 225 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000055 00000056 00000057 00000058 00000059 00000060 00000061 00000062 00000063 00000064 00000065 00000066 00000067 00000068 00000069 00000070 00000071 00000072 00000073 00000074 00000075 00000076 00000077 00000078 00000079 00000080 00000081 00000082 00000083 00000084 } } private void init_outputlayer() { _inputs[2] = new double[_numoutputs][]. for (int i = 0. i++) { _inputs[0][i][1] = 1. _weights[0][i][1] = 1. i < _hiddennodes.

_weights[2][i][_hiddennodes] = 1. } } private void init_hiddenweights() { for (int i = 0. i < _hiddennodes. i++) { _inputs[2][i][_hiddennodes] = 1. i++) { for (int j = 0. i < _numinputs. } } private void init_inputweights() { for (int i = 0. j < _numinputs. j++) { _weights[1][i][j] = Rand.0. } } } private void init_outputweights() Page 161 of 225 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000085 00000086 00000087 00000088 00000089 00000090 00000091 00000092 00000093 00000094 00000095 00000096 00000097 00000098 00000099 00000100 00000101 00000102 00000103 00000104 00000105 00000106 00000107 00000108 00000109 00000110 00000111 00000112 00000113 00000114 } //Output layer bias for (int i = 0.NextDouble().0.NextDouble(). i < _numoutputs. i++) { _weights[0][i][0] = Rand.

NextDouble(). i < _numinputs. i++) { for (int j = 0. } } } #endregion #region internal /********************\ |* INTERNAL *| \********************/ private double sigmoid(double input) { return (1 / (1 + Math. double sum = 0. try { //Calculate results for this layer for (int i = 0.Pow(Math. j < _hiddennodes. } private bool Runinput() { bool success = true. -input))).E. i++) { Page 162 of 225 .0. i < _numoutputs. j++) { _weights[2][i][j] = Rand.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000115 00000116 00000117 00000118 00000119 00000120 00000121 00000122 00000123 00000124 00000125 00000126 00000127 00000128 00000129 00000130 00000131 00000132 00000133 00000134 00000135 00000136 00000137 00000138 00000139 00000140 00000141 00000142 00000143 00000144 { for (int i = 0.

j++) { sum += (_inputs[0][i][j] * _weights[0][i][j]). try { //Feed forward the results from input layer for (int i = 0.Error).OK. i++) { for (int j = 0. "Error".William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000145 00000146 00000147 00000148 00000149 00000150 00000151 00000152 00000153 00000154 00000155 00000156 00000157 00000158 00000159 00000160 00000161 00000162 00000163 00000164 00000165 00000166 00000167 00000168 00000169 00000170 00000171 00000172 00000173 00000174 for (int j = 0. j < 2. Page 163 of 225 . sum = 0.Show("Error processing input layer". } return success. j++) { _inputs[1][i][j] = _outputs[0][j]. } } catch { MessageBox. j < _numinputs. double sum = 0. } _outputs[0][i] = sigmoid(sum). MessageBoxIcon.0.0. } private bool Runhidden() { bool success = true. i < _hiddennodes. success = false. MessageBoxButtons.

success = false.Error).0. i < _hiddennodes. MessageBoxIcon. j < _numinputs + 1. } _outputs[1][i] = sigmoid(sum).OK. MessageBoxButtons. sum = 0.0.Show("Error processing hidden layer". j++) { sum += (_inputs[1][i][j] * _weights[1][i][j]). } } catch { MessageBox. i++) { for (int j = 0.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000175 00000176 00000177 00000178 00000179 00000180 00000181 00000182 00000183 00000184 00000185 00000186 00000187 00000188 00000189 00000190 00000191 00000192 00000193 00000194 00000195 00000196 00000197 00000198 00000199 00000200 00000201 00000202 00000203 00000204 } } //Calculate results for this layer for (int i = 0. } private bool Runoutput() { bool success = true. try Page 164 of 225 . double sum = 0. } return success. "Error".

i < _numoutputs. success = false. i++) { for (int j = 0. sum = 0. j++) { _inputs[2][i][j] = _outputs[1][j]. "Error". MessageBoxButtons. i < _numoutputs.OK.Show("Error processing output layer".0. } _outputs[2][i] = sigmoid(sum).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000205 00000206 00000207 00000208 00000209 00000210 00000211 00000212 00000213 00000214 00000215 00000216 00000217 00000218 00000219 00000220 00000221 00000222 00000223 00000224 00000225 00000226 00000227 00000228 00000229 00000230 00000231 00000232 00000233 00000234 { //Feed forward the results from hidden layer for (int i = 0. } } catch { MessageBox. MessageBoxIcon. } Page 165 of 225 . j++) { sum += (_inputs[2][i][j] * _weights[2][i][j]). i++) { for (int j = 0. } } //Calculate results for this layer for (int i = 0. } return success. j < _hiddennodes + 1. j < _hiddennodes.Error).

MessageBoxIcon. success = false.Show("Error: Incorrect number of inputs supplied to NN".William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000235 00000236 00000237 00000238 00000239 00000240 00000241 00000242 00000243 00000244 00000245 00000246 00000247 00000248 00000249 00000250 00000251 00000252 00000253 00000254 00000255 00000256 00000257 00000258 00000259 00000260 00000261 00000262 00000263 00000264 #endregion #region external /********************\ |* EXTERNAL *| \********************/ public bool Run(double[] inputs) { bool success = true. MessageBoxButtons. if(success) success = Runhidden(). if(success) success = Runinput(). i++) //Each input node has only one real input //and a bias .OK.the number of nodes corresponds //to the number of inputs accepted. } if (success) for (int i = 0. "Incorrect number of inputs". _inputs[0][i][0] = inputs[i]. Page 166 of 225 .Error). //The numbers of inputs must match up if (inputs.Length != _numinputs) { MessageBox. i < _numinputs.

int numoutputs. int hiddennodes. } public double[] GetOutput() { //Return the outputs from the //output layer return _outputs[2]. int SetRandom) { //Set random number generator Rand = new Random(SetRandom). _numoutputs = numoutputs. //Set network structure descriptors _numinputs = numinputs. _hiddennodes = hiddennodes.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000265 00000266 00000267 00000268 00000269 00000270 00000271 00000272 00000273 00000274 00000275 00000276 00000277 00000278 00000279 00000280 00000281 00000282 00000283 00000284 00000285 00000286 00000287 00000288 00000289 00000290 00000291 00000292 00000293 00000294 if(success) success = Runoutput(). //We'll always have 3 layers Page 167 of 225 . return success. } #endregion #region constructor /********************\ |* CONSTRUCTOR *| \********************/ public Network(int numinputs.

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3
00000295 00000296 00000297 00000298 00000299 00000300 00000301 00000302 00000303 00000304 00000305 00000306 00000307 00000308 00000309 00000310 } //Input, Hidden, and Output _inputs = new double[3][][]; _weights = new double[3][][]; _outputs = new double[3][]; init_inputlayer(); init_hiddenlayer(); init_outputlayer(); init_bias(); init_inputweights(); init_hiddenweights(); init_outputweights(); } #endregion }

9.1.3.2 - “GANetwork.cs”
00000001 00000002 00000003 00000004 00000005 00000006 00000007 00000008 00000009 00000010 00000011 00000012 using using using using using using using System; System.Collections.Generic; System.Linq; System.Text; System.Windows.Forms; _05025397; System.Collections;

//This class contains Genetic Algorithm specific functions //in addition to the basic feed foward neural net functionality. namespace _05025397.Model

Page 168 of 225

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3
00000013 { 00000014 00000015 00000016 00000017 00000018 00000019 00000020 00000021 00000022 00000023 00000024 00000025 00000026 00000027 00000028 00000029 00000030 00000031 00000032 00000033 00000034 00000035 00000036 00000037 00000038 00000039 00000040 00000041 00000042

public class GANetwork : Network, IComparable { #region data /********************\ |* DATA *| \********************/ double _MSqE; #endregion #region .NET /********************\ |* .NET *| \********************/ //Interface_Implementation //Return Value Meanings: //-<Zero: x < y //-Zero: x == y //->Zero: x > y public int CompareTo(object a) { GANetwork b; if (a is GANetwork) { b = a as GANetwork; return _MSqE.CompareTo(b.MSqE); }

Page 169 of 225

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3
00000043 00000044 00000045 00000046 00000047 00000048 00000049 00000050 00000051 00000052 00000053 00000054 00000055 00000056 00000057 00000058 00000059 00000060 00000061 00000062 00000063 00000064 00000065 00000066 00000067 00000068 00000069 00000070 00000071 00000072 return 0; } #endregion #region external /********************\ |* EXTERNAL *| \********************/ public void getMsQE(double[][] inputs, double[][] outputs) { double sum = 0.0; int counter = 0; double[] netoutputs = new double[_numoutputs]; for (int i = 0; i < inputs.Length; i++) { base.Run(inputs[i]); netoutputs = base.GetOutput(); for (int j = 0; j < netoutputs.Length; j++) { sum += (outputs[i][j] - netoutputs[j]) * (outputs[i][j] - netoutputs[j]); counter++; } } _MSqE = (sum / counter); }

Page 170 of 225

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3
00000073 00000074 00000075 00000076 00000077 00000078 00000079 00000080 00000081 00000082 00000083 00000084 00000085 00000086 00000087 00000088 00000089 00000090 00000091 00000092 00000093 00000094 00000095 00000096 00000097 00000098 00000099 00000100 00000101 00000102 public double[] getWeights() { ArrayList collater = new ArrayList(); //Collect all the weights in an array list //then spit them out as a 1D array of doubles try { for (int i = 0; i < _weights.Length; i++) { for (int j = 0; j < _weights[i].Length; j++) { for (int k = 0; k < _weights[i][j].Length; k++) { collater.Add(_weights[i][j][k]); } } } } catch { MessageBox.Show("Fatal Error collating weights to 1D array", "Error", MessageBoxButtons.OK, MessageBoxIcon.Error); Application.Exit(); } return (double[])collater.ToArray(typeof(double)); } public void setWeights(double[] weights)

Page 171 of 225

j++) { for (int k = 0. i++) { for (int j = 0.Exit().Length. j < _weights[i]. } } } } catch { MessageBox. k++. "Error". k < _weights[i][j]. } } /********************\ |* GETTERS/SETTERS *| \********************/ public double MSqE { Page 172 of 225 . Application.Length. MessageBoxButtons.Length.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000103 00000104 00000105 00000106 00000107 00000108 00000109 00000110 00000111 00000112 00000113 00000114 00000115 00000116 00000117 00000118 00000119 00000120 00000121 00000122 00000123 00000124 00000125 00000126 00000127 00000128 00000129 00000130 00000131 00000132 { //Take a 1D array of doubles and apply //to the correct positions in our weights //array try { for (int i = 0. MessageBoxIcon. wc++) { _weights[i][j][k] = weights[wc]. wc = 0. i < _weights.Show("Fatal Error adding 1D weight array to 3D weight array".Error).OK.

Noutputs.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000133 00000134 00000135 00000136 00000137 00000138 00000139 00000140 00000141 00000142 00000143 00000144 00000145 00000146 00000147 00000148 00000149 00000150 00000151 } get { return _MSqE. int Nhnodes. } set { _MSqE = value. int SetRandom) : base(Ninputs. Nhnodes. } } #endregion #region constructor /********************\ |* CONSTRUCTOR *| \********************/ public GANetwork(int Ninputs.0. } #endregion } Page 173 of 225 . SetRandom) { MSqE = 0. int Noutputs.

namespace FANN_Wrapper { public ref class FANN_Cascade //Managed wrapper for UnManaged_FANN_CasC //which is a C++ interface to the C programmed //FANN dll. double TargetErr.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 9. using namespace System::Runtime::InteropServices.h" using namespace System. int OutputL.4 . public: /********************\ |* EXTERNAL *| \********************/ FANN_Cascade(int InputL.1.4. { public: /********************\ |* DATA *| \********************/ UnManaged_FANN_CasC* UMWrapper. Page 174 of 225 .1.h" #include "UnManaged_FANN_CasC.“FANN_Wrapper.1 . double LearnRate.DLL Wrapper for FANN functions 9.h” 00000001 00000002 00000003 00000004 00000005 00000006 00000007 00000008 00000009 00000010 00000011 00000012 00000013 00000014 00000015 00000016 00000017 00000018 00000019 00000020 00000021 00000022 00000023 00000024 00000025 00000026 #pragma once #include "UnManaged_FANN_BackProp.

public: /********************\ |* GETTERS/SETTERS *| \********************/ property String^ _te_output { String^ get() { return gcnew String((wchar_t *) UMWrapper->get_report_test()). } } property array<double>^ besterrorlist { array<double>^ get() MaxNeurons). } } property String^ _tr_output { String^ get() { return gcnew String((wchar_t *) UMWrapper->get_report_train()).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000027 00000028 00000029 00000030 00000031 00000032 00000033 00000034 00000035 00000036 00000037 00000038 00000039 00000040 00000041 00000042 00000043 00000044 00000045 00000046 00000047 00000048 00000049 00000050 00000051 00000052 00000053 00000054 00000055 00000056 int ReportInterval. bool Test(void). bool Train(void). Page 175 of 225 . int ~FANN_Cascade(void). !FANN_Cascade(void).

{ public: /********************\ |* DATA *| \********************/ UnManaged_FANN_Backprop* UMWrapper. const double *ar = UMWrapper->get_besterrs(size). public ref class FANN_BackProp //Managed wrapper for UnManaged_FANN_Backprop //which is a C++ interface to the C programmed //FANN dll. } return arout. array<double>^ arout = gcnew array<double>(size). } } }.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000057 00000058 00000059 00000060 00000061 00000062 00000063 00000064 00000065 00000066 00000067 00000068 00000069 00000070 00000071 00000072 00000073 00000074 00000075 00000076 00000077 00000078 00000079 00000080 00000081 00000082 00000083 00000084 00000085 00000086 { int size. public: Page 176 of 225 . i < size. i++) { arout[i] = ar[i]. for(int i = 0.

~FANN_BackProp(void). int OutputL. bool Train(void). } } property String^ _tr_output { String^ get() { return gcnew String((wchar_t *) UMWrapper->get_report_train()). int MaximumIteration). int HiddenL. public: /********************\ |* GETTERS/SETTERS *| \********************/ property String^ _te_output { String^ get() { return gcnew String((wchar_t *) UMWrapper->get_report_test()). double LearnRate. int ReportInterval. double TargetErr. bool Test(void).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000087 00000088 00000089 00000090 00000091 00000092 00000093 00000094 00000095 00000096 00000097 00000098 00000099 00000100 00000101 00000102 00000103 00000104 00000105 00000106 00000107 00000108 00000109 00000110 00000111 00000112 00000113 00000114 00000115 00000116 /********************\ |* EXTERNAL *| \********************/ FANN_BackProp(int InputL. !FANN_BackProp(void). } Page 177 of 225 .

Page 178 of 225 . } } }. } return arout. i++) { arout[i] = ar[i]. for(int i = 0. const double *ar = UMWrapper->get_besterrs(size).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000117 00000118 00000119 00000120 00000121 00000122 00000123 00000124 00000125 00000126 00000127 00000128 00000129 00000130 00000131 00000132 00000133 00000134 00000135 00000136 00000137 00000138 00000139 } } property array<double>^ besterrorlist { array<double>^ get() { int size. i < size. array<double>^ arout = gcnew array<double>(size).

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3

9.1.4.2 - “FANN_Wrapper.cpp”
00000001 00000002 00000003 00000004 00000005 00000006 00000007 00000008 00000009 00000010 00000011 00000012 00000013 00000014 00000015 00000016 00000017 00000018 00000019 00000020 00000021 00000022 00000023 00000024 00000025 00000026 00000027 00000028 00000029 #include "stdafx.h" #include "FANN_Wrapper.h" namespace FANN_Wrapper { /********************\ |* FANN_BackProp *| \********************/ //Constructor FANN_BackProp::FANN_BackProp(int InputL, int HiddenL, int OutputL, double LearnRate, double TargetErr, int ReportInterval, int MaximumIteration) { UMWrapper = new UnManaged_FANN_Backprop( InputL, HiddenL, OutputL, LearnRate, TargetErr, ReportInterval, MaximumIteration); } //Destructor FANN_BackProp::~FANN_BackProp(void) { if (UMWrapper) { delete UMWrapper; UMWrapper = 0; } }

Page 179 of 225

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3
00000030 00000031 00000032 00000033 00000034 00000035 00000036 00000037 00000038 00000039 00000040 00000041 00000042 00000043 00000044 00000045 00000046 00000047 00000048 00000049 00000050 00000051 00000052 00000053 00000054 00000055 00000056 00000057 00000058 00000059

//Finalizer FANN_BackProp::!FANN_BackProp(void) { if (UMWrapper) { delete UMWrapper; UMWrapper = 0; } }

//Train bool FANN_BackProp::Train(void) { return UMWrapper->Train(); } //Test bool FANN_BackProp::Test(void) { return UMWrapper->Test(); } /********************\ |* FANN_Cascade *| \********************/ //Constructor FANN_Cascade::FANN_Cascade(int InputL, int OutputL,

Page 180 of 225

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3
00000060 00000061 00000062 00000063 00000064 00000065 00000066 00000067 00000068 00000069 00000070 00000071 00000072 00000073 00000074 00000075 00000076 00000077 00000078 00000079 00000080 00000081 00000082 00000083 00000084 00000085 00000086 00000087 00000088 00000089 double LearnRate, double TargetErr, int ReportInterval, int MaxNeurons) { UMWrapper = new UnManaged_FANN_CasC( InputL, OutputL, LearnRate, TargetErr, ReportInterval, MaxNeurons); } //Destructor FANN_Cascade::~FANN_Cascade(void) { if (UMWrapper) { delete UMWrapper; UMWrapper = 0; } } //Finalizer FANN_Cascade::!FANN_Cascade(void) { if (UMWrapper) { delete UMWrapper; UMWrapper = 0; } } //Train

Page 181 of 225

William Sayers

05025397

Genetic Algorithms and Neural Networks
Milestone 3
00000090 00000091 00000092 00000093 00000094 00000095 00000096 00000097 00000098 00000099 00000100 } bool FANN_Cascade::Train(void) { return UMWrapper->Train(); } //Test bool FANN_Cascade::Test(void) { return UMWrapper->Test(); }

Page 182 of 225

4.h> #include <fann_cpp. double _Momentum. double _TargetErr.h> class UnManaged_FANN_Backprop //An interface to the C programmed //FANN dll. int _HiddenL.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 9. { private: /********************\ |* DATA *| \********************/ //Network Structure int _InputL. //Output for C# Page 183 of 225 . //Training Parameters double _LearnRate. int _OutputL.3 .1.“UnManaged_FANN_BackProp.h” 00000001 00000002 00000003 00000004 00000005 00000006 00000007 00000008 00000009 00000010 00000011 00000012 00000013 00000014 00000015 00000016 00000017 00000018 00000019 00000020 00000021 00000022 00000023 00000024 00000025 00000026 00000027 00000028 00000029 #pragma once #include <iostream> #include <string> #include <sstream> #include <vector> #include <doublefann. int _MaximumIteration. int _ReportInterval.

00000039 struct fann_train_data *data. 00000056 bool Test(void).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000030 std::wstringstream tr_output. 00000035 double *bsterr. 00000032 wchar_t *wchoutput. 00000031 std::wstringstream te_output. int MaximumIteration). 00000040 00000041 public: 00000042 /********************\ 00000043 |* EXTERNAL *| 00000044 \********************/ 00000045 //Constructor 00000046 UnManaged_FANN_Backprop(int InputL. 00000049 00000050 //Destructor 00000051 ~UnManaged_FANN_Backprop(void). 00000047 double LearnRate. int HiddenL. 00000057 00000058 public: 00000059 /********************\ Page 184 of 225 . int OutputL. 00000048 int ReportInterval. double TargetErr. 00000052 00000053 //Interface functions 00000054 //accessed from C# 00000055 bool Train(void). 00000036 00000037 //FANN Data 00000038 struct fann *ann. 00000033 00000034 std::vector<double> vecbsterr.

Page 185 of 225 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000060 00000061 00000062 00000063 00000064 00000065 00000066 }. const double* get_besterrs(int &size). const wchar_t* get_report_test(void). |* GETTERS/SETTERS *| \********************/ const wchar_t* get_report_train(void).

int MaximumIteration) { _InputL = InputL. bsterr = NULL. int OutputL.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 9.1. _MaximumIteration = MaximumIteration.h" #include "UnManaged_FANN_BackProp. _ReportInterval = ReportInterval.4. _LearnRate = LearnRate.cpp” 00000001 00000002 00000003 00000004 00000005 00000006 00000007 00000008 00000009 00000010 00000011 00000012 00000013 00000014 00000015 00000016 00000017 00000018 00000019 00000020 00000021 00000022 00000023 00000024 00000025 00000026 00000027 00000028 00000029 #include "StdAfx. _HiddenL = HiddenL. wchoutput = NULL. double LearnRate. _OutputL = OutputL. _TargetErr = TargetErr. } //Destructor UnManaged_FANN_Backprop::~UnManaged_FANN_Backprop() { fann_destroy_train(data).4 .h" //Constructor UnManaged_FANN_Backprop::UnManaged_FANN_Backprop (int InputL. int ReportInterval.“UnManaged_FANN_BackProp. ann = NULL. int HiddenL. double TargetErr. data = NULL. Page 186 of 225 .

_InputL.) { throw(""). _OutputL). success = false. wchoutput = NULL. int reportcounter = 0. try { data = fann_read_train_from_file("dlltrdata.. bool success = true. } //Train bool UnManaged_FANN_Backprop::Train(void) { static int firstrun = false. double error. delete wchoutput. } if (fann_num_input_train_data(data) != _InputL) Page 187 of 225 . ann = fann_create_standard(3. bsterr = NULL. _HiddenL.dat").. } catch(.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000030 00000031 00000032 00000033 00000034 00000035 00000036 00000037 00000038 00000039 00000040 00000041 00000042 00000043 00000044 00000045 00000046 00000047 00000048 00000049 00000050 00000051 00000052 00000053 00000054 00000055 00000056 00000057 00000058 00000059 delete bsterr.

//Set the learning rate fann_set_learning_rate(ann. (float) _LearnRate). //Standard backprop fann_set_training_algorithm(ann. FANN_TRAIN_BATCH). success = false. success = false. 1). //Same range the GA's weights are //initialised too Page 188 of 225 . } fann_set_activation_steepness_hidden(ann. fann_set_activation_function_output (ann. //Sigmoid Activation Functions (the same one //the GA uses).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000060 00000061 00000062 00000063 00000064 00000065 00000066 00000067 00000068 00000069 00000070 00000071 00000072 00000073 00000074 00000075 00000076 00000077 00000078 00000079 00000080 00000081 00000082 00000083 00000084 00000085 00000086 00000087 00000088 00000089 { throw(""). 1). FANN_SIGMOID). } if (fann_num_output_train_data(data) != _OutputL) { throw(""). fann_set_activation_function_hidden (ann. FANN_SIGMOID). fann_set_activation_steepness_output(ann.

\n". data). Page 189 of 225 . if (reportcounter == _ReportInterval) { tr_output << "Best error at iteration [ " << i << " ] was " << error << ".0).push_back(error).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000090 00000091 00000092 00000093 00000094 00000095 00000096 00000097 00000098 00000099 00000100 00000101 00000102 00000103 00000104 00000105 00000106 00000107 00000108 00000109 00000110 00000111 00000112 00000113 00000114 00000115 00000116 00000117 00000118 00000119 fann_randomize_weights(ann. vecbsterr. if (error < _TargetErr) { tr_output << "\nNetwork matching or improving upon target error" << "found at iteration [ " << i << " ]. i < _MaximumIteration.0. 0.". } } fann_destroy_train(data). } reportcounter++. //Training Loop for (int i = 0. 1. reportcounter = 0. i = _MaximumIteration + 1. i++) { //Train one epoch then check the //mean squared error error = (double) fann_train_epoch(ann.

dat"). double *error = NULL. for (int j = 0. j++) { Page 190 of 225 . } te_output << "\nOutput achieved was". 0. j < _OutputL. data = fann_read_train_from_file("dlltedata. i++) { calc_out = fann_run(ann. } //Test bool UnManaged_FANN_Backprop::Test(void) { fann_type *calc_out = NULL. 1). te_output << "\nWith inputs".William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000120 00000121 00000122 00000123 00000124 00000125 00000126 00000127 00000128 00000129 00000130 00000131 00000132 00000133 00000134 00000135 00000136 00000137 00000138 00000139 00000140 00000141 00000142 00000143 00000144 00000145 00000146 00000147 00000148 00000149 firstrun = true. data->input[i]). for (int j = 0. return true. j++) { te_output << " [" << data->input[i][j] << "] ". te_output << "\n\n\t\t\t~~Network Testing~~". for (unsigned int i = 0. j < _InputL. fann_scale_train_data(data. i < fann_length_train_data(data). error = new double[fann_length_train_data(data)].

tr_output.length() + 1]. te_output << fann_test_data(ann. j < _OutputL. data). delete calc_out. Page 191 of 225 . delete error. wchoutput = NULL.str().str(). } //get_report const wchar_t* UnManaged_FANN_Backprop::get_report_train(void) { if (wchoutput != NULL) { delete wchoutput.c_str()). return true. } wchoutput = new wchar_t[tr_output. wcscpy(wchoutput. } } te_output << "\nMean Squared Error with these inputs and outputs is:". } te_output << "\nOutput Desired was". j++) { te_output << " [" << data->output[i][j] << "] ". for (int j = 0.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000150 00000151 00000152 00000153 00000154 00000155 00000156 00000157 00000158 00000159 00000160 00000161 00000162 00000163 00000164 00000165 00000166 00000167 00000168 00000169 00000170 00000171 00000172 00000173 00000174 00000175 00000176 00000177 00000178 00000179 te_output << " [" << calc_out[j] << "] ".

size = vecbsterr.str(). } bsterr = new double[vecbsterr. return wchoutput.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000180 00000181 00000182 00000183 00000184 00000185 00000186 00000187 00000188 00000189 00000190 00000191 00000192 00000193 00000194 00000195 00000196 00000197 00000198 00000199 00000200 00000201 00000202 00000203 00000204 00000205 00000206 00000207 00000208 00000209 return wchoutput. } wchoutput = new wchar_t[te_output.length() + 1].c_str()). } //get_besterrs const double* UnManaged_FANN_Backprop::get_besterrs(int &size) { if (bsterr == NULL) { delete bsterr. te_output. wcscpy(wchoutput. Page 192 of 225 .str(). wchoutput = NULL. bsterr = NULL. } const wchar_t* UnManaged_FANN_Backprop::get_report_test(void) { if (wchoutput != NULL) { delete wchoutput.size()].size().

i < vecbsterr. 9.size().h” 00000001 00000002 00000003 00000004 00000005 00000006 00000007 00000008 00000009 00000010 00000011 00000012 00000013 00000014 00000015 00000016 00000017 00000018 #pragma once #include <iostream> #include <string> #include <sstream> #include <vector> #include <doublefann. } vecbsterr.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000210 00000211 00000212 00000213 00000214 00000215 00000216 00000217 00000218 00000219 } for (unsigned int i = 0.1. return bsterr. //Training Parameters Page 193 of 225 .5 .4.clear(). int _OutputL.“UnManaged_FANN_CasC.h> class UnManaged_FANN_CasC { public: /********************\ |* DATA *| \********************/ //Network Structure int _InputL. i++) { bsterr[i] = vecbsterr[i].

Page 194 of 225 . 00000026 std::wstringstream te_output. 00000031 00000032 //FANN Data 00000033 struct fann *ann.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000019 double _LearnRate. 00000035 00000036 public: 00000037 /********************\ 00000038 |* EXTERNAL *| 00000039 \********************/ 00000040 00000041 //Destructor 00000042 ~UnManaged_FANN_CasC(void). 00000047 double LearnRate. 00000030 double *bsterr. 00000048 int ReportInterval. 00000023 00000024 //Output for C# 00000025 static std::wstringstream tr_output. 00000020 double _TargetErr. int MaxNeurons). 00000028 00000029 static std::vector<double> vecbsterr. int OutputL. 00000043 00000044 //Constructor 00000045 UnManaged_FANN_CasC 00000046 (int InputL. 00000027 wchar_t *wchoutput. 00000021 int _ReportInterval. 00000022 int _MaxNeurons. double TargetErr. 00000034 struct fann_train_data *data.

Page 195 of 225 . struct fann_train_data *train. 00000057 unsigned int max_epochs. 00000066 00000067 const double* get_besterrs(int &size). 00000053 bool Test(void). 00000058 float desired_error. 00000059 00000060 public: 00000061 /********************\ 00000062 |* GETTERS/SETTERS *| 00000063 \********************/ 00000064 const wchar_t* get_report_train(void). 00000068 }.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000049 00000050 //Interface functions 00000051 //accessed from C# 00000052 bool Train(void). unsigned int epochs_between_reports. unsigned int epochs). 00000065 const wchar_t* get_report_test(void). 00000054 00000055 static int FANN_API Report_Callback 00000056 (struct fann *ann.

1. int ReportInterval.h" //Static variable declarations std::wstringstream UnManaged_FANN_CasC::tr_output. _LearnRate = LearnRate. int OutputL.h" #include "UnManaged_FANN_CasC.4. //Constructor UnManaged_FANN_CasC::UnManaged_FANN_CasC (int InputL. int MaxNeurons) { _InputL = InputL. ann = NULL. data = NULL.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 9.cpp” 00000001 00000002 00000003 00000004 00000005 00000006 00000007 00000008 00000009 00000010 00000011 00000012 00000013 00000014 00000015 00000016 00000017 00000018 00000019 00000020 00000021 00000022 00000023 00000024 00000025 00000026 00000027 00000028 00000029 #include "StdAfx. bsterr = NULL. double TargetErr. _TargetErr = TargetErr. wchoutput = NULL. std::vector<double> UnManaged_FANN_CasC::vecbsterr. } //Destructor UnManaged_FANN_CasC::~UnManaged_FANN_CasC(void) Page 196 of 225 . _OutputL = OutputL. _ReportInterval = ReportInterval. double LearnRate. _MaxNeurons = MaxNeurons.6 .“UnManaged_FANN_CasC.

William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000030 00000031 00000032 00000033 00000034 00000035 00000036 00000037 00000038 00000039 00000040 00000041 00000042 00000043 00000044 00000045 00000046 00000047 00000048 00000049 00000050 00000051 00000052 00000053 00000054 00000055 00000056 00000057 00000058 00000059 { delete bsterr. return wchoutput. } wchoutput = new wchar_t[tr_output. bsterr = NULL. } const wchar_t* UnManaged_FANN_CasC::get_report_test(void) { if (wchoutput != NULL) { delete wchoutput. wcscpy(wchoutput. delete wchoutput. fann_destroy_train(data). wchoutput = NULL. wchoutput = NULL. tr_output.str().length() + 1].str(). } //get_report const wchar_t* UnManaged_FANN_CasC::get_report_train(void) { if (wchoutput != NULL) { delete wchoutput. Page 197 of 225 .c_str()).

i++) { bsterr[i] = vecbsterr[i]. bsterr = NULL. } //get_besterrs const double* UnManaged_FANN_CasC::get_besterrs(int &size) { if (bsterr == NULL) { delete bsterr. for (unsigned int i = 0.size().size()]. } vecbsterr.str().clear().William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000060 00000061 00000062 00000063 00000064 00000065 00000066 00000067 00000068 00000069 00000070 00000071 00000072 00000073 00000074 00000075 00000076 00000077 00000078 00000079 00000080 00000081 00000082 00000083 00000084 00000085 00000086 00000087 00000088 00000089 wchoutput = NULL.size().c_str()). return wchoutput. return bsterr. i < vecbsterr. size = vecbsterr. wcscpy(wchoutput. } wchoutput = new wchar_t[te_output.length() + 1].str(). } bsterr = new double[vecbsterr. te_output. } Page 198 of 225 .

00000111 00000112 fann_set_learning_rate(ann. 00000100 } 00000101 catch(. 00000113 00000114 //Some more network customisation here 00000115 //might be nice in future.. (float)_LearnRate). 00000096 00000097 try 00000098 { 00000099 data = fann_read_train_from_file("dlltrdata.. 00000104 } 00000105 00000106 if (fann_num_input_train_data(data) != _InputL) 00000107 throw(""). Page 199 of 225 .) 00000102 { 00000103 throw(""). _OutputL). 2.dat"). 00000118 00000119 fann_set_quickprop_mu(ann. _InputL.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000090 //Train 00000091 bool UnManaged_FANN_CasC::Train() 00000092 { 00000093 int reportcounter = 0.0). 00000108 00000109 if (fann_num_output_train_data(data) != _OutputL) 00000110 throw(""). 0. 00000094 00000095 ann = fann_create_shortcut(2. 00000116 00000117 fann_set_quickprop_decay(ann.0).

00000133 00000134 fann_set_activation_function_output(ann. 00000127 00000128 fann_set_learning_rate(ann. 00000135 00000136 fann_randomize_weights(ann. 1). data. (float) _LearnRate).01f). _ReportInterval. 1). 0. 00000137 fann_set_cascade_output_change_fraction(ann. 00000124 00000125 fann_set_activation_steepness_output(ann.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000120 00000121 fann_set_cascade_weight_multiplier(ann.0. 1). FANN_SIGMOID). 00000131 00000132 fann_set_activation_function_hidden(ann. 00000142 00000143 fann_cascadetrain_on_data(ann. _MaxNeurons. 00000129 00000130 fann_set_training_algorithm(ann. 00000144 00000145 return true. 00000140 00000141 fann_set_callback(ann. 00000138 fann_set_cascade_candidate_change_fraction(ann. UnManaged_FANN_CasC::Report_Callback). FANN_SIGMOID). (float) _TargetErr). 1. FANN_STOPFUNC_MSE). 00000126 fann_set_activation_steepness_hidden(ann. 150). FANN_TRAIN_QUICKPROP). 00000146 } 00000147 00000148 //Callback for reporting 00000149 int FANN_API UnManaged_FANN_CasC::Report_Callback Page 200 of 225 .01f). 0. 00000122 00000123 fann_set_cascade_max_out_epochs(ann. 0.0). 00000139 fann_set_train_stop_function(ann.

\n". double *error = NULL.push_back(MSqE). unsigned int epochs_between_reports. unsigned int max_epochs. return 0. float desired_error.". struct fann_train_data *train. Page 201 of 225 . vecbsterr. fann_scale_train_data(data. static int node = 1. data = fann_read_train_from_file("dlltedata. } node++. 0.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000150 00000151 00000152 00000153 00000154 00000155 00000156 00000157 00000158 00000159 00000160 00000161 00000162 00000163 00000164 00000165 00000166 00000167 00000168 00000169 00000170 00000171 00000172 00000173 00000174 00000175 00000176 00000177 00000178 00000179 (struct fann *ann. } //Test bool UnManaged_FANN_CasC::Test() { fann_type *calc_out = NULL. tr_output << "Best error at node [ " << node << " ] was " << MSqE << ". if (MSqE <= desired_error) { tr_output << "\nNetwork matching or improving upon target error" << "found at node [ " << node << " ]. unsigned int epochs) { double MSqE = fann_get_MSE(ann).dat"). 1).

data). i++) { calc_out = fann_run(ann. for (int j = 0. j < _OutputL. i < fann_length_train_data(data). j < _OutputL. j++) { te_output << " [" << calc_out[j] << "] ". te_output << fann_test_data(ann. } te_output << "\nOutput achieved was:". j < _InputL. } te_output << "\nOutput Desired was:". for (int j = 0. for (int j = 0. NULL).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000180 00000181 00000182 00000183 00000184 00000185 00000186 00000187 00000188 00000189 00000190 00000191 00000192 00000193 00000194 00000195 00000196 00000197 00000198 00000199 00000200 00000201 00000202 00000203 00000204 00000205 00000206 00000207 00000208 00000209 fann_set_callback(ann. te_output << "\n\n\t\t\t~~Network Testing~~". for (unsigned int i = 0. error = new double[fann_length_train_data(data)]. Page 202 of 225 . te_output << "\nWith inputs:". j++) { te_output << " [" << data->output[i][j] << "] ". data->input[i]). } } te_output << "\nMean Squared Error with these inputs and outputs is:". j++) { te_output << " [" << data->input[i][j] << "] ".

Page 203 of 225 . delete calc_out.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 00000210 00000211 00000212 00000213 00000214 00000215 } delete error. return true.

UML Class Diagram The class diagram on the following pages represents the structure of the various classes within the program and how they are combined together to create a whole.2 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 9. Page 204 of 225 .

William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Page 205 of 225 .

William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Page 206 of 225 .

mit.edu/ga/ o Accessed: 31/10/2008 Page 207 of 225 .Libraries Researched • GALib o Matthew Wall o http://lancet.3 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 -BackProp Previous Page -CascadeCorrelation 1 FANN_Cascade -UMWrapper +FANN_Cascade(in AllDataToFillAttributes) +Train() : bool +Test() : bool 1 -UMWrapper 1 FANN_BackProp +FANN_BackProp(in AllDataToFillAttributes) +Train() : bool +Test() : bool 1 1 1 UnManaged_FANN_BackProp -_InputL : int -_HiddenL : int -_OutputL : int -_LearnRate : double -_TargetErr : double -_ReportInterval : int -_MaximumIteration : int -_Momentum : double -tr_output -te_output -wchoutput -vecbsterr -bsterr -ann -data +UnManaged_FANN_Backprop(in AddDataToFillAttributes) +~UnManaged_FANN_Backprop() +Train() : bool +Test() : bool UnManaged_FANN_CasC -_InputL : int -_OutputL : int -_LearnRate : double -_TargetErr : double -_ReportInterval : int -_MaxNeurons : int -tr_output -te_output -wchoutput -vecbsterr -bsterr -ann -data +UnManaged_FANN_CasC(in AllDataToAFillAttributes) +~UnManaged_FANN_CasC() +Train() : bool +Test() : bool +Report_Callback(in Attributes) : int 1 1 1 1 FANN_Library 9.

edu/au/~daniel/software/libneural/ o Accessed: 31/10/2008 • Libann o Anonymous o http://www.org/libann/index.cimne.com/flood/ o Accessed: 31/10/2008 • Annie o Asim Shankar o http://annie.html o Accessed: 31/10/2008 • Flood o Roberto Lopez o http://www.net/ o Accessed: 31/10/2008 • FANN (Fast Artificial Neural Network Library) o Steffen Nissen Page 208 of 225 . Franklin o http://ieee.uow.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 • Libneural o D.nongnu.sourceforge.

ac.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 o http://leenissen.edu/stat/data/Fisher.xhtml Virus Classification o http://www.uah.math.ox.html Page 209 of 225 .uk/pub/PRNN/README.4 .stats.php o Accessed: 31/10/2008 9.Datasets • XOR 1 0 1 0 1 1 1 1 0 0 0 0 • • Fishers Iris Data o http://www.dk/fann/index.

Objectives FINAL YEAR DEGREE PROJECT Project Objectives & Deliverables • • • • • • Research the fields of genetic algorithms and neural networks. and a document describing its implementation and usage. evaluate training methods for neural networks. Page 210 of 225 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 9. Deliverable three: A document detailing and explaining the evaluation of different training techniques. Build an application that is capable of training a neural network with a genetic algorithm and with a backpropagation system. • • • Deliverable one: A research document describing and detailing the research done for objectives one and two. and the conclusions derived from that evaluation. Deliverable two: An application meeting the requirements of objectives three and four. Using above application. and in which cases these methods are suitable. Research current information on training neural networks using genetic algorithms. and that allows manual adjustment of relevant variables. Build a user-interface for the aforementioned application allowing for easy alteration of appropriate variables.5 . Identify the most effective ways of training neural networks using genetic algorithms. as well as previous research.

ps Fahlman. C. Wiley Professional Computing. J. Carnegie Mellon: http://www2.ps.edu/user/sef/www/publications/cascor-tr.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 10 . Neural Network in C++. Curved trajectory prediction using a self-organizing neural network. (1992). 2009. Davis. 2009.unm.cs. The Cascade-Correlation Learning Architecture. from http://www. S. (1991). Branke. G. AI: Neural Network for Beginners pt. Chapter 14: Neural Networks.com/UploadFile/scottlysle/PassData01142007013005AM/PassData. S.cmu. 2009. from C# Corner: http://www. from School of Computer Science. M. (1995). from Codeproject: www.ps Lysle.cmu. Bourg.csharpcorner. (2006). January 16). E. An Empirical Study of Learning Speed in Back-Propagation Networks. E. A. D. or the preservation of favoured races in the struggle for life. Retrieved April 16. 356-358). M.aspx Blum. Passing Data between Windows Forms.Works Cited Barber. (2001). & Seeman. Graphics processing unit computation of neural networks. from http://www.edu/user/sef/www/publications/qp-tr. C. Retrieved April 13.edu/~chris2d/papers/CED_thesis.pdf Fahlman. Retrieved April 16. Retrieved April 16.cs.cs. Carnegie Mellon: http://www-2. (1859). 2009. & Lebiere. Retrieved April 13. Bourg. (1988). 2009.edu/afs/cs. from School of Computer Science. S. (2007.com/KB/recipes/NeuralNetwork_1. S... On the origin of the species by means of natural selection. 2009. E.codeproject. & G. 1. (2004).gz Darwin. C.cmu. In D. London: John Murray.aifb.edu/afs/cs.unikarlsruhe. Retrieved April 13. O'Reilly.aspx Page 211 of 225 . Seeman.cmu. AI for Game Developers (pp.de/~jbr/Papers/GaNNOverview.

Learning Logic.html Marshall. (n. A. J. from Computer Science .gz Matthews.. S. from MSDN: http://msdn. & Srikanth. Retrieved 21 01. (1943).generation5.d. from Artificial Neural Networks: http://www. 2009.aspx Mitchell.aspx Microsoft. from generation5: http://www.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Marczyk. Stanford University. Using C++ Interop (Implicit PInvoke).UNC-Chapel Hill: http://www. (2002).microsoft. Massachusetts: MIT Press. V. D. Retrieved April 16. Genetic algorithms and evolutionary computation. Microsoft. Bulletin of Mathematical Biophysics . 115-133. Office of Technology Licensing. Invention Report S81-64.talkorigins.learnartificialneuralnetworks. 386-408. Robot Control.microsoft. Retrieved April 16. from MSDN: http://msdn.componentmodel.asp McCulloch. 2009.backgroundworker. M.com/robotcontrol. Curved trajectory prediction using a selforganizing neural network. (1958). (2004). Retrieved April 15. & Pitts.com/en-us/library/system. Page 212 of 225 .dk/fann/ Parker. An introduction to genetic algorithms. B. Psychological Review . 2009. (2008). from FANN: http://leenissen.htm Rosenblatt.. F.ps.edu/~marshall/WWW/PAPERS/curve9912. 65. J.com/en-us/library/2x8kf7zx. The perceptron: A probabilistic theory for information storage and organization in the brain. Back-propagation for the Uninitiated. A logical calculus of the ideas immanent in nervous activity. (1996). A. (2009). 2009. W. 2009. Retrieved April 13. Nissen. Retrieved April 15. (2009). W. Retrieved April 15. (1999). 2009.com/faqs/genalg/genalg. (1982). BackgroundWorker Class.cs. 2009. 5.).unc.org/content/2002/bp. from http://www.

doc. Beyond Regression: New tools for prediction and analysis in the behavioral sciences.php?title=Main_Page Page 213 of 225 . & Dimitrios. C. (1974). PhD. from Imperial College London: http://www. S. Harvard University. P.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Stergiou.d. Retrieved 21 01. (1997).org/wiki/index. ZedGraph.html#Contents Werbos. 2009. 2009. Retrieved April 15.ac.). Neural Networks.ic..uk/~nd/surprise_96/journal/vol4/cs11/report. (n. from http://zedgraph.

cs. Bourg.pdf DiLascia. Bourg. 2009. from http://www.codeproject.aifb. (2006). Retrieved April 13. Curved trajectory prediction using a self-organizing neural network. On the origin of the species by means of natural selection. & Seeman. (2007.aspx Blum. AI: Neural Network for Beginners pt. S. August 6). A. from MSDN: http://msdn. Retrieved April 13. 2009. AI for Game Developers (pp.aspx Barber. 2009.ps.com/KB/cs/GA_ANN_XOR. 2009. (1859). AI: Neural Network for Beginners pt.codeproject. 1. Retrieved April 13.codeguru. Retrieved April 13. 356-358).unikarlsruhe. 2009. Chapter 14: Neural Networks.edu/~chris2d/papers/CED_thesis. 2009. In D.aspx Barber. Seeman. M. Retrieved April 13. London: John Murray. D. 2. P.microsoft. Branke.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 11 . (1995). S. Call Unmanaged DLLs from C#. or the preservation of favoured races in the struggle for life.de/~jbr/Papers/GaNNOverview. from Codeguru: http://www. J. from http://www. from Codeproject: http://www. S.. (2006). (2004).aspx DLL Tutorial for Beginners. Neural Network in C++. (2002).php/c9855/ Page 214 of 225 . G.com/KB/recipes/NeuralNetwork_1. AI: Neural Network for Beginners pt. (2006). C. M.codeproject.com/enus/magazine/cc301501.com/KB/recipes/Backprop_ANN.gz Darwin. Graphics processing unit computation of neural networks. (1992). Retrieved April 13.com/Cpp/Cpp/cpp_mfc/tutorials/article. Killing Processes Cleanly. Wiley Professional Computing. (2001). 3. & G. Davis. Retrieved April 13.unm. O'Reilly. from Codeproject: www. 2009.Bibliography Barber. C. from Codeproject: http://www. E.

2009.ps Fahlman.edu/user/sef/www/publications/cascor-tr. A. Carnegie Mellon: http://www-2. Genetic algorithms and evolutionary computation.html Marshall. 2009.generation5.cs.. from generation5: http://www. W.cmu. E.edu/user/sef/www/publications/qp-tr. Retrieved April 16. 2009.UNC-Chapel Hill: http://www. J.edu/~marshall/WWW/PAPERS/curve9912. (n.gz Matthews.d.ps GoogleChartSharp. An Empirical Study of Learning Speed in Back-Propagation Networks. J. Bulletin of Mathematical Biophysics .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Fahlman.cs. J. from generation5: http://www. S. Passing Data between Windows Forms. Retrieved April 13. 2009. 2009.com/UploadFile/scottlysle/PassData01142007013005AM/PassData. 2009. Back-propagation for the Uninitiated. & Srikanth. (1999). 2009.com/faqs/genalg/genalg. 115-133. from School of Computer Science.cs.org/content/2000/cbpnet.). (2007.ps. January 16). A logical calculus of the ideas immanent in nervous activity. Curved trajectory prediction using a selforganizing neural network. S. Retrieved 21 01. (2002)..cmu. J.asp Matthews.generation5.cmu. S. Retrieved April 15.edu/afs/cs. (1988).cmu.com/p/googlechartsharp/ Lysle. Retrieved April 15.). 2009. from Computer Science . C. V.asp Matthews. from http://code. from School of Computer Science. W.asp McCulloch.aspx Marczyk. 2009.org/content/2000/xorga. (1991). Back-Propagation: CBPNet. Retrieved 26 01. & Lebiere. (2000).org/content/2002/bp. Retrieved April 16. (1943). E. Page 215 of 225 .. (2004). Carnegie Mellon: http://www2. from C# Corner: http://www. & Pitts. from http://www.edu/afs/cs. Retrieved April 15. generation5. A.csharpcorner. Retrieved April 16.talkorigins. 5. from XORGA: http://www.google.generation5.d. (n.unc. The Cascade-Correlation Learning Architecture.

com/en-us/library/system. Nguyen.dk/fann/ Nissen. Retrieved 21 01.componentmodel.d.NET. M.). Neural Networks. from MSDN: http://msdn.dk/fann/fann_1_2_0/r1837. Massachusetts: MIT Press.). B. S.d. BackgroundWorker Class.). (n. 2009. 2009.html Nissen. from FANN: http://leenissen.html Nissen. (2009). (1996).aspx Mitchell. (2009). An introduction to genetic algorithms. (n.dk/fann/html/files/fann_data-h. Retrieved 21 01. D. Improving the learning speed of 2-layer neural networks by choosinginitial values of the adaptive weights. Retrieved 21 01. Office of Technology Licensing. (n. Retrieved April 16. D.d..aspx Microsoft.org/nplot/wiki/ Parker.htm Page 216 of 225 .d..microsoft. from Artificial Neural Networks: http://www. Retrieved 21 01. Nissen. from FANN: http://leenissen. (n. & Widrow. struct fann_train_data. 2009.dk/fann/html/files/fann-h.html Nplot Charting Library for .microsoft.html#fann_train_enum Nissen.d.dk/fann/fann_1_2_0/r1597. Retrieved April 16. S. S.learnartificialneuralnetworks. Stanford University. 2009. Robot Control. 2009. (1990). S. 2009. Retrieved 26 01.com/robotcontrol. 2009. Reference Manual. from MSDN: http://msdn. Invention Report S81-64. Retrieved 21 01. from Datatypes: http://leenissen. 21-26. from FANN: http://leenissen. Using C++ Interop (Implicit PInvoke).). S. (n.). 2009.d. (1982).William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Microsoft. (n. 2009. struct fann. from http://netcontrols. 1990 IJCNN International Joint Conference on .).backgroundworker. 1990.com/en-us/library/2x8kf7zx. Retrieved April 15. (2008). B. from FANN: http://leenissen. Learning Logic.

1991)....A feed-forward neural network .............. Beyond Regression: New tools for prediction and analysis in the behavioral sciences..................d.......... The perceptron: A probabilistic theory for information storage and organization in the brain........... 65....... 386-408.....1 ........Tables 12.....A human neuron ...................... Retrieved April 15..An artificial neuron . ZedGraph.......... 35 Page 217 of 225 .......The initial state of a cascade correlation neural network...... Neural Networks. Stergiou............... from http://zedgraph..... (n............... 13 Figure 4 . ...... 30 Figure 6 ...................... S..William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Rosenblatt............. Retrieved 21 01..... P...... 12 Figure 3 ......................... The Cascade-Correlation Learning Architecture...............The sigmoid function shape ...php?title=Main_Page 12 .....ic........ 10 Figure 2 ................ 2009............html#Contents Werbos........... PhD.......... Psychological Review ..................Table of Figures Figure 1 .................... C.........org/wiki/index................................... F....................).......... (1974).local optimum (units are undefined in this case)...ac..... & Dimitrios......... The circular connections are adjustable weights (Fahlman & Lebiere... Harvard University.......... 2009...... 27 Figure 5 ........... (1958)........... from Imperial College London: http://www...... ..uk/~nd/surprise_96/journal/vol4/cs11/report...... (1997).doc.................

........................................................Network Settings tab...... 46 Figure 19 ............................. The CascadeCorrelation Learning Architecture........... 49 Figure 22 ....Mean squared error tab design (Main form) . 43 Figure 13 .......................................Learning Algorithm tab ................... with two nodes added (Fahlman & Lebiere..........................The second state of a cascade correlation neural network................................................. The Cascade-Correlation Learning Architecture........................................................ 51 Page 218 of 225 ....Working form design (Working form) ........................ ............. 44 Figure 15 ... 1991)......................................Dataset selection tab design (Wizard form)................. 36 Figure 9 – Network and output tab design (Main form) ........ 48 Figure 21 .............................................................Mean squared error graph tab ..........................The third state of a cascade correlation neural network...................About form design (About form) ............................................... 35 Figure 8 ................................................... The square connections are locked weights(Fahlman & Lebiere...........Licenses display design (About form) ..............................Design of the training algorithm tab (Wizard form) ............................................ 42 Figure 12 ............ 42 Figure 11 ................................... 45 Figure 17 ..................Dataset display design (About form) ......................... 45 Figure 16 ............. 1991)........ 47 Figure 20 ................William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Figure 7 ..........Network and Output tab....................................................Design of the network settings tab (Wizard form) .................................Dataset tab design (Main form) .............The dataset tab ........... 50 Figure 23 .............................................. 46 Figure 18 .......................... 41 Figure 10 ......... A first node has been added................... 44 Figure 14 ................

.................................................Number of iterations to achieve target (Viruses) . 61 Figure 30 ................. 52 Figure 26 ........2 ..................Working form .......William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Figure 24 .Mean squared error upon last training test (Iris data) ............... 54 Figure 28 .............Dataset tab .............................................................................About form...The mean squared error at testing (Iris data) ........... 62 Figure 32 ......... 51 Figure 25 ............................Summation of the inputs and weights ............................ 68 12...................Mean squared error upon last training test (XOR) ........... 64 Figure 33.............. main view .About form........................................... 67 Figure 37 ............................... 21 Page 219 of 225 .......A sigmoid function(Bourg & Seeman................... 65 Figure 35 ..... 67 Figure 36 ............. 12 Equation 2 ........... Licenses view .......................Number of iterations to achieve target (XOR) ........................................ .................... 54 Figure 29 ..................... 2002) ............................... 2004).................The mean squared error at testing (XOR)...................... 64 Figure 34.... 53 Figure 27 .........About form................. 61 Figure 31 ...........................................Mean squared error at testing (Viruses).................. 13 Equation 3 – The delta rule(Matthews....Mean squared error upon last training test (Viruses)....................................... as defined in the graph above........ Datasets view ...... Back-propagation for the Uninitiated..............................................Number of iterations to achieve target (Iris data) ...........................Table of Equations Equation 1 ........... In this example “y” is the function output and “t” is the function input........................................................

..........The partial derivative of ’s’ (Fahlman & Lebiere........... 'p' is the training pattern............................................................. 2002)......................... 21 Equation 7 ................................. 59 Page 220 of 225 .............The partial derivative of ‘s’ with respect to each of the candidates incoming weights ‘wi’................ Back-propagation for the Uninitiated....................Table of tables Table 1 ........... ... 21 Equation 6 ............. 2002) ...............Genetic Algorithm testing data (XOR) ....................................... The Cascade-Correlation Learning Architecture............................ 2002) ........................................... the quantities ‘`v ’ and ‘`Eo’ are the values of v and Eo averaged over all patterns (Fahlman & Lebiere..p’ is the input the candidate receives from unit ‘i’ for pattern ‘p’.......... 1991).......differentiated sigmoid function (Matthews......... Backpropagation for the Uninitiated.............. 22 Equation 8 ...................................... Backpropagation for the Uninitiated...........The statistical approach to deciding relevant data .................... where ‘oo’ is the sign of the correlation between the candidates value output ‘o’.............................................Back-propagation testing data (XOR).... 59 Table 3 . 2002).Altered delta calculation rule for the output layer(Matthews...... ‘f'p’ is the derivative for pattern 'p' of the candidate unit's activation function with respect to the sum of it’s inputs and ‘Ii... 33 Equation 9 .................... ...............William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Equation 4 ............The definition of 's' given that 'o' is the network output at which the error is measured....... 1991)..................................................................................... 15 Table 2 ...3 ....................................................Altered perceptron learning rule (Matthews............ 34 Equation 10 ...........................................................Altered delta calculation rule for the hidden layer(s) (Matthews...... 34 12...................... 'v' is the candidate unit's value................................. The CascadeCorrelation Learning Architecture.......................... 21 Equation 5 ...................... Back-propagation for the Uninitiated.................

.......... 60 Table 5 .....................................Cascade correlation testing data (Iris data) ....Back-propagation testing data (Iris data) ............Genetic Algorithms testing data (Viruses) ........... 66 Table 10 .................Back-propagation testing data (Viruses) .................................. 63 Table 7 ... 66 Page 221 of 225 ....................................... 63 Table 8 ............................................................................................................................................... 63 Table 6 ...........Genetic algorithm testing data (Iris data) ...William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 Table 4 ..................................Cascade Correlation testing data (XOR) .................. 66 Table 9 .........Cascade Correlation testing data (Viruses) .......

back-propagation network was running and learning XOR.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 13 . o Decided on XOR as a starting point. o Decision made to adjust the weights of the neural network during training as opposed to using a genetic algorithm to establish network structure. however. • 10/10/2008 o GALib decided as a genetic algorithm starting point o Further non-web references as a goal • • 16/10/2008 o Investigation into various neural network libraries taking place. o Submitted objectives. o Agreed on further investigation into GALib and genetic algorithms.Diary • 01/10/2008 – Research progress satisfactory. • 31/10/2008 o First presentation complete and ready o References building up o Project plan complete • 10/11/2008 o Presentation went well o Progress satisfactory • 14/11/2008 o Discussed document formatting and layout Page 222 of 225 . 17/10/2008 o Investigating more deeply into journals based on research already conducted. • 24/10/2008 o University computers were down so code demo was impossible.

William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 o Progress satisfactory. o Progress satisfactory • 20/02/2009 o Progress satisfactory Page 223 of 225 . • • • • • • • • • 19/11/2008 o Milestone one submitted successfully. 13/02/2009 o Have decided to extend the project by including cascade correlation learning algorithm and comparison to genetic algorithms. 21/11/2008 o Progress satisfactory 28/11/2008 o Progress satisfactory 05/12/2008 o Progress satisfactory 12/12/2008 o Progress satisfactory 09/01/2009 o Progress satisfactory 16/01/2009 o Progress satisfactory 23/01/2009 o Progress satisfactory 30/01/2009 o Milestone two has been successfully submitted o Started investigating other areas to explore outside of project requirements. • • 06/02/2009 o Still exploring extra areas for project to cover.

o Report satisfactory • 05/04/2009 o Further work on user interface needed o Finish C# genetic algorithm code (C++ code does work but is very slow) o Report satisfactory • 06/04/2009 o C# genetic algorithm code is complete. is much faster (fixed several bugs along the way) and retains none of the problematic memory leaks. have decided to implement in managed language (C#) to try to solve this o Report continues satisfactorily • 04/04/2009 o To do: About box Add dataset notes Decided on virus classification as a third dataset via the custom dataset functionality. Page 224 of 225 . o Have started report • 02/04/2009 o Main problems experienced with genetic algorithm seem related to memory leaks.William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 • • • • 27/02/2009 o Progress satisfactory 06/03/2009 o Project progress delayed by other coursework 13/03/2009 o Project progress delayed by other coursework 20/03/2009 o Project progressing. catch-up time for previous two weeks will be used over Easter break.

o Report satisfactory. Page 225 of 225 .William Sayers 05025397 Genetic Algorithms and Neural Networks Milestone 3 o Report satisfactory • 10/04/2009 o Fully functional project – commencing report.

Sign up to vote on this title
UsefulNot useful