KANEKO Tomoyuki

kaneko@graco.c.u-tokyo.ac.jp
[Japanese|English]

He is an associate professor at the Department of "Multi-Disciplinary Sciences [Systems Sciences]" at Graduate School of the University of Tokyo and is a member of graco research group. His interest is machine learning in games.

Research

• Open source Shogi program and its library:
  • large-scale learning of its evaluation function (#parameters is about 19 millions in total, about 1.4 millions with non-zero values). (submitted)
  • parallel depth-first proof number search (AAAI2010)
  • distributed game-tree search. (in progress)
• try 3x4shogi at 81dojo: three bots 81{NOIR,BLANC,GRIS} will play better if you achive more wins against them.
• past interests: automated feature construction for evaluation functions of general game players