Flexible and Stable Pattern Generation by Evolving Constrained Plastic Neurocontrollers
Autor: | Patrick Henaff, Cecilia Tapia Siles, Thierry Hoinville |
---|---|
Přispěvatelé: | Instituto Italiano di Tecnologia (IIT), Ministero dell'Istruzione-Ministero dell'Economia e Finanze, Equipes Traitement de l'Information et Systèmes (ETIS - UMR 8051), Ecole Nationale Supérieure de l'Electronique et de ses Applications (ENSEA)-Centre National de la Recherche Scientifique (CNRS)-CY Cergy Paris Université (CY) |
Jazyk: | angličtina |
Rok vydání: | 2011 |
Předmět: |
Computer science
Stability (learning theory) Evolutionary robotics Experimental and Cognitive Psychology 03 medical and health sciences Behavioral Neuroscience adaptive synapses 0302 clinical medicine Control theory Homeostasis [INFO.INFO-RB]Computer Science [cs]/Robotics [cs.RO] legged locomotion 030304 developmental biology Flexibility (engineering) 0303 health sciences Artificial neural network Quantitative Biology::Neurons and Cognition business.industry Robustness (evolution) Evolvability Hebbian theory Artificial intelligence business 030217 neurology & neurosurgery plastic neural networks evolutionary robotics |
Zdroj: | Adaptive Behavior Adaptive Behavior, SAGE Publications, 2011, 19, pp.187-207. ⟨10.1177/1059712311403631⟩ |
ISSN: | 1059-7123 |
Popis: | International audience; In evolutionary robotics, plastic neural network models proved to be promising for evolving adaptive behaviors. In particular, neurocontrollers incorporating hebbian synapses have been shown to be useful for implementing conflicting sub-behaviors. Numerous interesting complex tasks assume such flexibility. However, those evolved controllers often exhibit behavioral instability, as simulation time is extended beyond the short limit used during evolution. In this paper, we propose constrained plastic models inspired by neural homeostasis phenomena, in order to evolve flexible and stable pattern generators for single-legged locomotion. Comparative results show that constrained controllers perform better than unconstrained ones in both terms of evolvability and behavioral stability. Functional analyses of the best evolved controller unveil the adaptivity, robustness and homeostasis arising from the statically constrained plasticity. Interestingly, homeostasis evolved implicitly without relying on any active homeostatic mechanisms and is implemented through hebbian plasticity, usually considered destabilizing |
Databáze: | OpenAIRE |
Externí odkaz: |