Timezone: »

 
Oral
ELF OpenGo: an analysis and open reimplementation of AlphaZero
Yuandong Tian · Jerry Ma · Qucheng Gong · Shubho Sengupta · Zhuoyuan Chen · James Pinkerton · Larry Zitnick

Tue Jun 11 11:00 AM -- 11:20 AM (PDT) @ Hall B
in Deep RL 1 »

The AlphaGo, AlphaGo Zero, and AlphaZero series of algorithms are remarkable demonstrations of deep reinforcement learning's capabilities, achieving superhuman performance in the complex game of Go with progressively increasing autonomy. However, many obstacles remain in the understanding of and usability of these promising approaches by the research community. Toward elucidating unresolved mysteries and facilitating future research, we propose ELF OpenGo, an open-source reimplementation of the AlphaZero algorithm. ELF OpenGo is the first open-source Go AI to convincingly demonstrate superhuman performance with a perfect (20:0) record against global top professionals. We apply ELF OpenGo to conduct extensive ablation studies, and to identify and analyze numerous interesting phenomena in both the model training and in the gameplay inference procedures. Our code, models, selfplay datasets, and auxiliary data are publicly available.

[ Video] 

Author Information

Yuandong Tian (Facebook AI Research)
Jerry Ma (Facebook AI Research)
Qucheng Gong (Facebook AI Research)
Shubho Sengupta (Facebook AI Research)
Zhuoyuan Chen (Facebook)

Zhuoyuan Chen is a researcher in Facebook AI Research. Before joining Facebook, he worked at Baidu Research USA as a research scientist. Zhuoyuan graduated from Northwestern University. Zhuoyuan's research interest mainly focuses on reinforcement learning, meta learning and computer vision.

James Pinkerton (Facebook AI Research)
Larry Zitnick (Facebook AI Research)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors