Evolving Artificial Neural Networks through Evolutionary Programming.
Xin Yao, Yong Liu · 1996
Artificial neural network (ANN) architecture design has been one of the most tedious and difficult tasks in ANN applications due to the lack of satisfactory and systematic methods of designing a near optimal architecture. Evolutionary algorithms have been shown to be very effective in evolving novel ANN architectures for various problems. This paper proposes a new automatic method for simultaneously evolving ANN architectures and weights. The method has been applied to four realworld data sets in the medical domain and achieved very good results. 1 Introduction Artificial neural networks (ANNs) have been used widely in many application areas in recent years. Most applications use feed-forward ANNs and the back-propagation (BP) training algorithm. There are numerous variants of the classical BP algorithm and other training algorithms, but these algorithms assume a fixed ANN architecture. They only train connection weights (including biases) in a fixed architecture that includes both co...