Repeated Inverse Reinforcement Learning
This addresses the challenge of improving human-agent collaboration in sequential decision-making by reducing unexpected behaviors, though it appears incremental as it builds on existing inverse reinforcement learning concepts.
The paper tackles the problem of an agent acting on behalf of a human across multiple tasks, aiming to minimize surprises by learning from demonstrations when suboptimal actions occur, and provides foundational formalizations and results for this repeated inverse reinforcement learning setup.
We introduce a novel repeated Inverse Reinforcement Learning problem: the agent has to act on behalf of a human in a sequence of tasks and wishes to minimize the number of tasks that it surprises the human by acting suboptimally with respect to how the human would have acted. Each time the human is surprised, the agent is provided a demonstration of the desired behavior by the human. We formalize this problem, including how the sequence of tasks is chosen, in a few different ways and provide some foundational results.