Neural network performance analysis using hanning window function as dynamic learning rate

In human brain the neurons are excited in a dynamic way. The response of different neurons varies widely because of the variation of electrical signal in every neuron. Backpropagation(BP) is a training algorithm where the learning of the Neural Network (NN) is done by a constant learning rate (LR). But to replicate the human brain function, the learning rate should be changed as the excitation of different neurons. In this paper a new learning algorithm is proposed called Hanning Window Neural Network (HWNN) to train the network. Here the window function is used to make the learning rate dynamic called Hanning learning rate (HLR) and for this dynamic learning rate the neural network outperforms than the existing BP algorithm. HWNN is extensively tested on five real world benchmark classification problems such as ionosphere, australian credit card, time series, wine and soybean identification. The proposed HWNN outperforms the existing BP in terms of generalization ability and also convergence rate.