Artificial Neural Network



). etc. 2 .Definition & Area of Application Neural Networks (NN): have remarkable ability to drive meaning from complicated or imprecise data.  exploratory analysis (data mining). voice.  “knowledge acquisition tools” that learn from examples Neural Networks are used for: pattern recognition (objects in images.  predictive models and control. medical diagnostics for diseases.

3 .Neural Networks  Inspired by natural decision making structures (real nervous systems and brains)  If you connect lots of simple decision making pieces together. they can make more complex decisions  Compose simple functions to produce complex functions  Neural networks:  Take multiple numeric input variables  Produce multiple numeric output values  Normally threshold outputs to turn them into discrete values.

The key element of this paradigm is the novel structure of the information processing system. process information. It is composed of a large number of highly interconnected processing elements (neurons) working in union to solve specific problems.What is an artificial neural network? An Artificial Neural Network (ANN) is an information processing paradigm that is inspired by the way biological nervous systems. such as the brain. 4 .

A network can consists of few to a few billion neurons connected in an array of different methods. ANN’s attempt to model these biological structures both in architectures and operations 5 .Cont. the tiny cell are brain is comprised of . The term Neural network is in fact a biological term which is collection of neuron.

It can be implemented without any problem. it can continue without any problem by their parallel nature. A neural network learns and does not need to be reprogrammed. Requires high processing time for large neural networks. 6 . When an element of the neural network fails. Disadvantages: The neural network needs training to operate.Advantage and Disadvantage Advantages:  A neural network can perform tasks that a linear program can      not.

the output of any layer does not affect that same layer.Neural Network Architectures  Feed-forward networks:Feed-forward ANNs allow signals to travel one way only. Feed-forward ANNs tend to be straight forward networks that associate inputs with outputs. There is no feedback (loops) i. from input to output. 7 .e. They are extensively used in pattern recognition.

Feedback architectures are also referred to as interactive or recurrent. their 'state' is changing continuously until they reach an equilibrium point. Feedback networks are very powerful and can get extremely complicated.  Recurrent /Feedback networks:Feedback networks can have signals travelling in both directions by introducing loops in the network. 8 . although the latter term is often used to denote feedback connections in single-layer organizations. They remain at the equilibrium point until the input changes and a new equilibrium needs to be found.Cont. Feedback networks are dynamic.

9 .Learning/Training of artificial neural networks A neural network has to be configured such that the application of a set of inputs produces (either 'direct' or via a relaxation process) the desired set of outputs. Various methods to set the strengths of the connections exist. Another way is to 'train' the neural network by feeding it teaching patterns and letting it change its weights according to some learning rule. One way is to set the weights explicitly. using a priori knowledge.

Cont. or by the system which contains the neural network (self-supervised). These input-output pairs can be provided by an external teacher. These are:  Supervised learning or Associative learning in which the network is trained by providing it with input and matching output patterns. We can categorize the learning situations in two distinct sorts. 10 .

11 . rather the system must develop its own representation of the input stimuli.  Unlike the supervised learning paradigm.  Unsupervised learning or Self-organization in which an (output) unit is trained to respond to clusters of pattern within the input. there is no a priori set of categories into which the patterns are to be classified.Cont. In this paradigm the system is supposed to discover statistically salient features of the input population.

Generally.Cont. following which there will be no more changes in its parameters. 12 . The self-organizing neural learning may be categorized under this type of learning.  The learning system grades its action good (rewarding) or bad (punishable) based on the environmental response and accordingly adjusts its parameters. parameter adjustment is continued until an equilibrium state occurs. Here the learning machine does some action on the environment and gets a feedback response from the environment.  Reinforcement Learning This type of learning may be considered as an intermediate form of the above two types of learning.

Application  Neural Network applicability to real world business     problem like sales forecasting. marketing. credit assignment evaluation etc Modeling and diagnosing the cardiovascular system Application in biomedical researches n medicine and instant physician Robotics Application Electronic Noses 13 . customer research.

7.9433.4.7335..90790 889022:89/0.0789./2 9070834.943  &30908:507..039 10.9:708419035:9545:.99073939035:9 3 985.990738.30/94708543/94..3.80/0.8.7.:89078415.574780941 .3 4:95:9  47$01 :39897..733 47$01 47.989.80/0.439 &38:507.0459843705708039.8810/7./2908890288:55480/94/8.904708394.905.943419035:9892:  .7094-0.

7085438017429003./0898.7332.3/098.7..733 07090 0.733   ..7.306:-7:289.3390720/. 100/-./:8920398 .209078 %0801 -0342470.30/4088420.90470/:3/0798950410.9070 -0342470..7. 0.-4.438/070/ ..904.209078 0307.7432039.80/439003..943439003.3083985.7/3 47-.7432039.-0.20907.7332.9014724190.-0.209078 %0801 47.94344/ 70.:78 1443.439 439 #03147./ 5:38.8../:898985.4393:0/:39.7.7432039 %00.3/..020390...733889027.3083985.70854380. 5..3330:7.733 %8950410.47/3 .0949508410.-0 -.7332.

:.47/-:83088 574-0208....3958.94309.7.7080.788902 55.55.:.8..3/ 389.893 ..943 0:7.88320390.:8942077080.943 0..55..  2.7093 .-99470.3 #4-49..081470.30. 4/03.4808  ..08320/.9433-420/.348390.70/9.0947.9743.7/4.855..3//.7.

Sign up to vote on this title
UsefulNot useful