Exploration-Exploitation in Reinforcement Learning: Function Approximation and Deep RL🔗
Abstract🔗
One of the major challenges in reinforcement learning (RL) is the trade-off between exploration of the environment to gather information and exploitation of the samples observed so far to execute "good" (nearly optimal) actions. In this seminar, we review how exploration techniques are paired with function approximation in continuous state-action spaces. In particular, we will focus on the integration of exploration mechanisms with deep learning techniques. The seminar should provide enough theoretical and algorithmic background to understand existing techniques and possibly devise novel methods. Throughout the talk, we will discuss open problems and possible future research directions.