Application of a Scaled Conjugate Gradient Algorithm for Feed Forward Artificial Neural Networks

摘要

This paper presents an implementation of a Scaled Conjugate Gradient (SCG) training algorithm for feed forward artificial neural networks (ANNs). Some background to ANNs is presented, training is described and the SCG algorithm developed. An application of SCG to training a Universal Window Filter (UWF) is presented and its performance compared to standard back propagation.