Reinforcement Learning via Reasoning from Demonstration
This addresses the challenge of sparse-reward tasks in reinforcement learning, offering a human-inspired approach that could improve agent learning efficiency, though it appears incremental in its current implementation.
The paper tackles the problem of sparse-reward reinforcement learning by proposing a framework where agents use demonstrations to develop causal models and reason from them, showing that a basic implementation is effective in various sparse-reward tasks.
Demonstration is an appealing way for humans to provide assistance to reinforcement-learning agents. Most approaches in this area view demonstrations primarily as sources of behavioral bias. But in sparse-reward tasks, humans seem to treat demonstrations more as sources of causal knowledge. This paper proposes a framework for agents that benefit from demonstration in this human-inspired way. In this framework, agents develop causal models through observation, and reason from this knowledge to decompose tasks for effective reinforcement learning. Experimental results show that a basic implementation of Reasoning from Demonstration (RfD) is effective in a range of sparse-reward tasks.