Optimization of Number of Hidden Neurons in Neural Networks

Sarika Shukla, Jameel Ahamad

Abstract: This paper carries study of Neural Network and analyses the effects of number of neurons in the hidden layer of the network on the output. This paper also contains study of conventional optimization methods for ascertaining the number of hidden neurons and analyses of neural network performance. Implementation of optimization criterion based on estimation of Signal to Noise Ratio Figure, Sum Squared Error and Delta Value method for ascertaining the number of hidden neurons and analyses of neural network performance. Comparison of results by using the methods as mentioned in the above points is presented here.

Keywords: Neural Network, Signal to Noise Ratio Figure, Sum Squared Error, Delta Value method