RT Journal Article
SR Electronic
T1 Human-level performance in 3D multiplayer games with population-based reinforcement learning
JF Science
JO Science
FD American Association for the Advancement of Science
SP 859
OP 865
DO 10.1126/science.aau6249
VO 364
IS 6443
A1 Jaderberg, Max
A1 Czarnecki, Wojciech M.
A1 Dunning, Iain
A1 Marris, Luke
A1 Lever, Guy
A1 Castañeda, Antonio Garcia
A1 Beattie, Charles
A1 Rabinowitz, Neil C.
A1 Morcos, Ari S.
A1 Ruderman, Avraham
A1 Sonnerat, Nicolas
A1 Green, Tim
A1 Deason, Louise
A1 Leibo, Joel Z.
A1 Silver, David
A1 Hassabis, Demis
A1 Kavukcuoglu, Koray
A1 Graepel, Thore
YR 2019
UL http://science.sciencemag.org/content/364/6443/859.abstract
AB Artificially intelligent agents are getting better and better at two-player games, but most real-world endeavors require teamwork. Jaderberg et al. designed a computer program that excels at playing the video game Quake III Arena in Capture the Flag mode, where two multiplayer teams compete in capturing the flags of the opposing team. The agents were trained by playing thousands of games, gradually learning successful strategies not unlike those favored by their human counterparts. Computer agents competed successfully against humans even when their reaction times were slowed to match those of humans.Science, this issue p. 859Reinforcement learning (RL) has shown great success in increasingly complex single-agent environments and two-player turn-based games. However, the real world contains multiple agents, each learning and acting independently to cooperate and compete with other agents. We used a tournament-style evaluation to demonstrate that an agent can achieve human-level performance in a three-dimensional multiplayer first-person video game, Quake III Arena in Capture the Flag mode, using only pixels and game points scored as input. We used a two-tier optimization process in which a population of independent RL agents are trained concurrently from thousands of parallel matches on randomly generated environments. Each agent learns its own internal reward signal and rich representation of the world. These results indicate the great potential of multiagent reinforcement learning for artificial intelligence research.