Question

1. What is the difference in the output layer between a neural network used for classification, and one used for regression?

2. Describe why we need to use regularization in neural networks.

Answer 1

`Hey,

Note: Brother in case of any queries, just comment in box I would be very happy to assist all your queries

The difference between a classification and regression is that a classification outputs a prediction probability for class/classes and regression provides a value. Hence, the final layer in neural network for classification predicts probability for classes whereas the final layer in neural network for regression provides a value.

However, we can make a neural network to output a value by simply changing the activation function in the final layer to output the values.

By changing the activation function such as sigmoid,relu,tanh,etc. we can use a function (f(x)=x. So while back propagation simply derive f(x).

For illustration I will provide you the forward and back ward pass for a single layer neural network regression below:

forward pass: inputs−>x

weightsinputtohidden−>w1

weightshiddentooutput−>w2

z2=w1∗x

a2=sigmoid(z2)

z3=w2∗a2

a3=f(z3)

backward pass:

targets−>Y

f(x)=x−>f′(x)=1

sigmoid′(x)−>sigmoid(x)(1−sigmoid(x))

d3=Y−a3

d2=w2∗d3

w2′=a2∗d3

w1′=d2∗a2′∗x

Here the d3 and d2 are layer wise errors.

Relapse is a factual measure utilized in back, contributing and different controls that endeavors to decide the quality of the connection between one ward variable (for the most part signified by Y) and a progression of other changing factors

The distinction between a grouping and relapse is that a characterization yields an expectation likelihood for class/classes and relapse gives an esteem. Thus, the last layer in neural system for grouping predicts likelihood for classes while the last layer in neural system for relapse gives an esteem.

Be that as it may, we can make a neural system to yield an incentive by just changing the actuation work in the last layer to yield the qualities.

By changing the actuation capacity, for example, sigmoid,relu,tanh,etc. we can utilize a capacity (f(x)=x. So while back engendering essentially determine f(x)

Kindly revert for any queries

Thanks.