Quick Propagation

An heuristic modification of the back propagation algorithm which attempts to use a simple quadratic model of the error surface (calculated independently along each weight) to speed convergence (Fahlman, 1988; Patterson, 1996). See, Neural Networks .