Discovery of backpropagation learning rules using geneticprogramming

摘要

The backpropagation learning rule is widespread computationalmethod for training multilayer networks. Unfortunately, backpropagationsuffers from several problems. The authors have used genetic programming(GP) to overcome some of these problems and to discover new supervisedlearning algorithms. A set of such learning algorithms has been comparedwith the standard backpropagation (SBP) learning algorithm on differentproblems and has been shown to provide better performances. The studyindicates that there exist many supervised learning algorithms betterthan, but similar to, SEP and that GP can be used to discover them