Conservative Exploration in Reinforcement Learning
This addresses safety concerns in real-world RL applications by providing guarantees on policy performance during learning.
The paper tackles the problem of ensuring that intermediate policies in reinforcement learning meet a minimum performance baseline during exploration, introducing conservative exploration for average reward and finite horizon problems with algorithms that guarantee constraint satisfaction and maintain learning ability.
While learning in an unknown Markov Decision Process (MDP), an agent should trade off exploration to discover new information about the MDP, and exploitation of the current knowledge to maximize the reward. Although the agent will eventually learn a good or optimal policy, there is no guarantee on the quality of the intermediate policies. This lack of control is undesired in real-world applications where a minimum requirement is that the executed policies are guaranteed to perform at least as well as an existing baseline. In this paper, we introduce the notion of conservative exploration for average reward and finite horizon problems. We present two optimistic algorithms that guarantee (w.h.p.) that the conservative constraint is never violated during learning. We derive regret bounds showing that being conservative does not hinder the learning ability of these algorithms.