Reinforcement Learning for Collaboration in Modern Video Games

Microsoft Research and a partner studio are announcing a new research collaboration that explores the potential of new AI techniques, such as deep reinforcement learning, in gaming. This talk demos their latest results in a team-based game, which is a perfect testing ground for reinforcement learning agents trained to collaborate. The talk gives a high-level technical overview and provides tips and tricks for making reinforcement learning work in gaming applications.

Date:

Regardez suivant