TY - GEN
T1 - Exploration conscious reinforcement learning revisited
AU - Shani, Lior
AU - Efroni, Yonathan
AU - Mannor, Shie
N1 - Publisher Copyright: © 2019 International Machine Learning Society (IMLS).
PY - 2019
Y1 - 2019
N2 - The Exploration-Exploitation tradeoff arises in Reinforcement Learning when one cannot tell if a policy is optimal. Then, there is a constant need to explore new actions instead of exploiting past experience. In practice, it is common to resolve the tradeoff by using a fixed exploration mechanism, such as ϵ-greedy exploration or by adding Gaussian noise, while still trying to learn an optimal policy. In this work, we take a different approach and study exploration-conscious criteria, that result in optimal policies with respect to the exploration mechanism. Solving these criteria, as we establish, amounts to solving a surrogate Markov Decision Process. We continue and analyze properties of exploration-conscious optimal policies and characterize two general approaches to solve such criteria. Building on the approaches, we apply simple changes in existing tabular and deep Reinforcement Learning algorithms and empirically demonstrate superior performance relatively to their non-exploration-conscious counterparts, both for discrete and continuous action spaces.
AB - The Exploration-Exploitation tradeoff arises in Reinforcement Learning when one cannot tell if a policy is optimal. Then, there is a constant need to explore new actions instead of exploiting past experience. In practice, it is common to resolve the tradeoff by using a fixed exploration mechanism, such as ϵ-greedy exploration or by adding Gaussian noise, while still trying to learn an optimal policy. In this work, we take a different approach and study exploration-conscious criteria, that result in optimal policies with respect to the exploration mechanism. Solving these criteria, as we establish, amounts to solving a surrogate Markov Decision Process. We continue and analyze properties of exploration-conscious optimal policies and characterize two general approaches to solve such criteria. Building on the approaches, we apply simple changes in existing tabular and deep Reinforcement Learning algorithms and empirically demonstrate superior performance relatively to their non-exploration-conscious counterparts, both for discrete and continuous action spaces.
UR - http://www.scopus.com/inward/record.url?scp=85078260861&partnerID=8YFLogxK
M3 - منشور من مؤتمر
T3 - 36th International Conference on Machine Learning, ICML 2019
SP - 9986
EP - 10012
BT - 36th International Conference on Machine Learning, ICML 2019
T2 - 36th International Conference on Machine Learning, ICML 2019
Y2 - 9 June 2019 through 15 June 2019
ER -