Multi-Agent Deep RL

Centralized Training with Decentralized Execution (CTDE)

Approach where agents are trained using centralized global information but execute their policies in a decentralized manner with local observations.

← Terug