Goal Misgeneralization in Deep Reinforcement Learning. (arXiv:2105.14111v3 [cs.LG] UPDATED)

Web: http://arxiv.org/abs/2105.14111

We study goal misgeneralization, a type of out-of-distribution generalization
failure in reinforcement learning (RL). Goal misgeneralization failures occur
when an RL agent retains its capabilities out-of-distribution yet pursues the
wrong goal. For instance, an agent might continue to competently avoid
obstacles, but navigate to the wrong place. In contrast, previous works have
typically focused on capability generalization failures, where an agent fails
to do anything sensible at test time. We formalize this distinction between
capability and goal generalization, provide the …

arxiv deep learning lg reinforcement reinforcement learning