1 paper
Alvaro Serra-Gomez, Thomas Moerland
Effective exploration in high-dimensional continuous control remains a central challenge in reinforcement learning. Planning-based methods address this by combining online planning…