Heuristically-accelerated reinforcement learning: A comparative analysis of performance
Nenhuma Miniatura disponível
Citações na Scopus
2
Tipo de produção
Artigo de evento
Data
2014
Autores
MARTINS, M. F.
Reinaldo Bianchi
Reinaldo Bianchi
Orientador
Periódico
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Título da Revista
ISSN da Revista
Título de Volume
Citação
MARTINS, M. F.; BIANCHI, R. Heuristically-accelerated reinforcement learning: A comparative analysis of performance. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), p. 15-27, 2014.
Texto completo (DOI)
Palavras-chave
Resumo
This paper presents a comparative analysis of three Reinforcement Learning algorithms (Q-learning, Q(λ)-learning and QS-learning) and their heuristically-accelerated variants (HAQL, HAQ(λ) and HAQS) where heuristics bias action selection, thus speeding up the learning. The experiments were performed in a simulated robot soccer environment which reproduces the conditions of a real competition league environment. The results clearly demonstrate that the use of heuristics substantially improves the performance of the learning algorithms. © 2014 Springer-Verlag.