Question

in training deep learning we have hidden layers how can i select it based on what ?

Answer 1

The number of hidden neurons should be in between the size of the input layer and size of output layers.

The number of hidden neurons should be less than the twice of the input layer size.

The number of hidden neurons can be calculated based on input layer and output layers size.

The number of hidden neurons should be equal to 2/3 rd of size of input layers + size of output layers.

Example:-

Let size of input layer=a

size of output layer =b

Then hidden layers= 2/3* a +b

The hidden layers are required if and only if the data must be separated non-linearly.

The optimal number of hidden units could easily be smaller than the number of inputs.

Selection of number of hidden layers :-

(No of inputs + No of outputs)^0.5 + n ( where n is the integer values range from 0 to 10), use trial and error method and find the optimal no of hidden layer neurons for the minimum value of mean squared error.

Generally one hidden layer is enough for the large majority of the problems to solve.

selection of number of hidden neurons in the hidden layer is based on above rules which we discussed in the starting.

If you have any queries regarding this comment below.