Momentum term heals the performance of Back Propagation Algorithm for digit recognition

Abstract

The objective of this research paper is to analyze the momentum term for the performance of Back Propagation (BP) Algorithm of Artificial Neural Network (ANN) for the problem of hand written digit recognition. The two variations of BP algorithm are simple BP algorithm and BP algorithm with momentum. In this paper BP with momentum is used for the performance analysis of BP algorithm. Numbers of parameters like training iterations, learning rate, neurons in the hidden layer are used during this analysis. The simulation results show that momentum term heals the performance of BP algorithm for hand written digit recognition application.