PC-PG: Policy Cover Directed Exploration for Provable Policy Gradient Learning

Aug-15-2025, 08:54:17 GMT–Neural Information Processing Systems 

In other words, the assumptions in these works imply that the state space is already well-explored.