TrajDeleter: Enabling Trajectory Forgetting in Offline Reinforcement Learning Agents

arXiv:2404.12530v1 Announce Type: cross
Abstract: Reinforcement learning (RL) trains an agent from experiences interacting with the environment. In scenarios where online interactions are impractical, offline RL, which trains the agent using pre-collected datasets, has become popular. While this new paradigm presents remarkable effectiveness across various real-world domains, like healthcare and energy management, there is a growing demand to enable agents to rapidly and completely eliminate the influence of specific trajectories from both the training dataset and the trained agents. To …

agent agents arxiv cs.cr cs.lg datasets domains energy environment experiences healthcare offline paradigm popular real trains trajectory world