You are on page 1of 2

Proposed Methods :

In doing project experimentation with implementation we have run into many distinct
models and after the result obtained by each model we not only do its comparison with an
earlier model, but also find out the cons which can be fulfilled by adding extra machinery .
Keeping in mind the earlier one model , its outcomes and disabilities we use our new model
which manages to acquire and correct the swindles of the model designed just before this .
Similar processors have been repeated with this model too and we are running into design
model some after the other in a way to solve the problem caused by the model just before
the presently designed model . With respect to the above phenomenon only , we reached to
design many more models which includes Expert System , MP Neuron Model , Perceptron
Model , Sigmoid Model , DeepNeural Network.

Expert System : It is the very first system that was developed to solve real world related
problems. It is basically based on or working on the decision making ability of humans. On
this model we predict the output on the basis of past available data or past experience. It is
helpful in some tasks like checking if the dengue disease is present or not in people and also
helps to decide LBW or not in a cricket match. But it also has some drawbacks i.e. it is very
hard to deal with large amounts of data. In an expert system we generally write a few lines of
code on the basis of required output. It also fails to give output where output is based on
emotions like if a company wants to hire some people for a specific post and interviewer select
and like a person because of his honesty and loyalty so this type of decision making task can
not be done by the expert system.

Advantages:
● Allows you to get quick and precise responses.
● Provides a dependable response to a recurring problem.
● It has the ability to accumulate and utilize scarce skills.

Limitations:
● Limitations in information can result in bad decisions.

● The cost of maintaining an expert system is unreasonably high.


● Because every situation is different, a human expert's answer may be
different andmore innovative as well.

McCulloch-Pitts Neuron (MP Neuron Model) : This model was proposed by the
neuroscientist Walter Pits . As being in depth knowledge about the human brain, he is
willing to design a model which can work having its capacity equivalent to the human
brain. Keeping in mind all the model related requirements he started working in, its model
makes use of different elements of Machine Learning than those in expert systems. In MP
Neuron we took input of only binary numbers as data i.e. (1,0). While doing tasks in MP
Neuron, we classify them first. Then the model we have used is named as Linear
Threshold Gate Model, it doesn’t set down a fixed value rather we predict the value here
put down in an algorithm and check the result accuracy. Conclusively this model
evaluates the Loss Entropy Error and obtains accuracy percentage in the result from it.
Six elements of Machine Learning for MP Neuron are as follows:
Data : The MP neuron receives binary input and outputs binary data. If the input data
isn't binary, it can be compressed before being fed to the model.
Classification : The categorization is also binary, meaning it is either 0 or 1. Based on the
inputand the threshold, the model can respond with a yes or no.
Model : It is made up of a single-parameter function. The data is compiled. A threshold
value is set. If the function's value is equal to or larger than threshold level, the output is
positive, and vice versa. It draws a graph between positive values that lie above the line
and negative values that lie below the line.
Loss Function : It is the difference between the predicted value and the actual value in
squareformat i.e. square loss function is applied .

You might also like