Provable Rich Observation Reinforcement Learning with Combinatorial Latent States
international conference on learning representations, 2021.
We introduce a problem setup and a provable reinforcement learning algorithm for rich-observation problems with latent combinatorially large state space.
We propose a novel setting for reinforcement learning that combines two common real-world difficulties: presence of observations (such as camera images) and factored states (such as location of objects). In our setting, the agent receives observations generated stochastically from a \"latent\" factored state. These observations are \"rich...More
PPT (Upload PPT)