MBEANN: Mutation-Based Evolving Artificial Neural Networks (CROSBI ID 37212)
Prilog u knjizi | izvorni znanstveni rad
Podaci o odgovornosti
Ohkura, Kazuhiro ; Yasuda, Toshiyuki ; Kawamatsu, Yuichi ; Matsumura, Yoshiyuki ; Ueda, Kanji
engleski
MBEANN: Mutation-Based Evolving Artificial Neural Networks
A novel approach to topology and weight evolving artifi - cial neural networks (TWEANNs) is presented. Compared with previous TWEANNs, this method has two major characteristics. First, a set of genetic operations may be designed without recombination because it often generates an off spring whose fi tness value is considerably worse than its parents. Instead, two topological mutations whose eff ect on fi tness value is assumed to be nearly neutral are provided in the genetic operations set. Second, a new encoding technique is introduced to defi ne a string as a set of substrings called operons. To examine our approach, computer simulations were conducted using the standard reinforcement learning problem known as the double pole balancing without velocity information. The results obtained were compared with NEAT results, which is recognised as one of the most powerful techniques in TWEANNs. It was found that our proposed approach yields competitive results, especially when the problem is diffi cult.
Multi-robot System, Reinforcement Learning, Autonomous
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
Podaci o prilogu
936-945.
objavljeno
Podaci o knjizi
Advances in Artificial Life
F. Almeida e Costa et al.
Berlin : Heidelberg: Springer
2007.
978-3-540-74912-7