DECAF: Deep Case-based Policy Inference for knowledge transfer in Reinforcement Learning

Autor(es)

Glatt, Ruben ; Da Silva, Felipe Leno ; da Costa Bianchi, Reinaldo Augusto ; Reali Costa, Anna Helena

• Tipo de documento: Artigo Científico
• Fonte: EXPERT SYSTEMS WITH APPLICATIONS; v. 156, p. 13-pg., 2020-10-15.

Resumo

Having the ability to solve increasingly complex problems using Reinforcement Learning (RL) has prompted researchers to start developing a greater interest in systematic approaches to retain and reuse knowledge over a variety of tasks. With Case-based Reasoning (CBR) there exists a general methodology that provides a framework for knowledge transfer which has been underrepresented in the RL literature so far. We formulate a terminology for the CBR framework targeted towards RL researchers with the goal of facilitating communication between the respective research communities. Based on this framework, we propose the Deep Case-based Policy Inference (DECAF) algorithm to accelerate learning by building a library of cases and reusing them if they are similar to a new task when training a new policy. DECAF guides the training by dynamically selecting and blending policies according to their usefulness for the current target task, reusing previously learned policies for a more effective exploration but still enabling the adaptation to particularities of the new task. We show an empirical evaluation in the Atari game playing domain depicting the benefits of our algorithm with regards to sample efficiency, robustness against negative transfer, and performance increase when compared to state-of-the-art methods. (C) 2020 Elsevier Ltd. All rights reserved. (AU)

• Processo FAPESP: 16/18792-9 - Descrição, representação e solução de jogos espaciais
• Beneficiário: Paulo Eduardo Santos
• Modalidade de apoio: Auxílio à Pesquisa - Parceria para Inovação Tecnológica - PITE
• Processo FAPESP: 16/21047-3 - ALIS: Aprendizado Autônomo em Sistemas Inteligentes
• Beneficiário: Anna Helena Reali Costa
• Modalidade de apoio: Auxílio à Pesquisa - Regular
• Processo FAPESP: 18/00344-5 - Reusando soluções de tarefas prévias em aprendizado por reforço multiagente
• Beneficiário: Felipe Leno da Silva
• Modalidade de apoio: Bolsas no Exterior - Estágio de Pesquisa - Doutorado
• Processo FAPESP: 15/16310-4 - Transferência de Conhecimento no Aprendizado por Reforço em Sistemas Multiagentes
• Beneficiário: Felipe Leno da Silva
• Modalidade de apoio: Bolsas no Brasil - Doutorado