Difference between revisions of "Kei Takada"
GerdIsenberg (talk | contribs) |
GerdIsenberg (talk | contribs) |
||
| Line 7: | Line 7: | ||
==Ezo== | ==Ezo== | ||
| − | Along with [[Masaya Honjo]], [[Hiroyuki Iizuka]], and [[Masahito Yamamoto]], Kei Takada is author of the [[Hex]] playing program [https://www.game-ai-forum.org/icga-tournaments/program.php?id=755 Ezo], which played the [[17th Computer Olympiad|17th Computer Olympiad]] in [https://en.wikipedia.org/wiki/Yokohama Yokohama] 2013, and the [[18th Computer Olympiad]] in [[Leiden University|Leiden]] 2015. Ezo uses 2-ply [[Alpha-Beta|alpha-beta]] search with an [[Evaluation | + | Along with [[Masaya Honjo]], [[Hiroyuki Iizuka]], and [[Masahito Yamamoto]], Kei Takada is author of the [[Hex]] playing program [https://www.game-ai-forum.org/icga-tournaments/program.php?id=755 Ezo], which played the [[17th Computer Olympiad|17th Computer Olympiad]] in [https://en.wikipedia.org/wiki/Yokohama Yokohama] 2013, and the [[18th Computer Olympiad]] in [[Leiden University|Leiden]] 2015. Ezo uses 2-ply [[Alpha-Beta|alpha-beta]] search with an [[Evaluation Function|evaluation function]] based on the theory of [https://en.wikipedia.org/wiki/Complex_network complex networks]. The function combines a betweenness measure of the player’s two sides with a shortest path metric that accounts for the degree of important nodes, further elaborted by Kei Takada at the [[Advances in Computer Games 14]] conference in Leiden 2015 <ref> [[Kei Takada]], [[Masaya Honjo]], [[Hiroyuki Iizuka]], [[Masahito Yamamoto]] ('''2015'''). ''[http://link.springer.com/chapter/10.1007%2F978-3-319-27992-3_21 Developing Computer Hex using Global and Local Evaluation based on Board Network Characteristics]''. [[Advances in Computer Games 14]], [https://acg2015.files.wordpress.com/2015/07/takada.pdf slides as pdf]</ref>. EZO switches to a [[Vadim Anshelevich#Hexy|Hexy-style]] evaluation when remaining time becomes short <ref>[[Ryan Hayward]], [[Broderick Arneson]], [[Shih-Chieh Huang]], [[Jakub Pawlewicz]] ('''2013'''). ''MOHEX Wins Hex Tournament''. [[ICGA Journal#36_3|ICGA Journal, Vol. 36, No. 3]], [https://webdocs.cs.ualberta.ca/~hayward/papers/rptYokohama.pdf pdf]</ref>. Ezo-CNN, which played the [[20th Computer Olympiad#Hex|20th Computer Olympiad 2017]], uses a [[Neural Networks#Convolutional|convolutional neural policy network]] for [[Move Ordering|move ordering]] <ref>[[Ryan Hayward]], [[Noah Weninger]] ('''2017'''). ''Hex 2017: MoHex wins the 11x11 and 13x13 tournaments''. [[ICGA Journal#39_34|ICGA Journal, Vol. 39, Nos. 3-4]]</ref> <ref> [[Kei Takada]], [[Hiroyuki Iizuka]], [[Masahito Yamamoto]] ('''2018'''). ''[https://link.springer.com/chapter/10.1007%2F978-3-319-75931-9_2 Computer Hex Algorithm Using a Move Evaluation Method Based on a Convolutional Neural Network]''. [https://link.springer.com/bookseries/7899 Communications in Computer and Information Science]</ref>. |
=Selected Publications= | =Selected Publications= | ||
Revision as of 13:49, 28 May 2018
Kei Takada [1]
Kei Takada,
a Japanese computer scientist affiliated with the Graduate School of Information Science and Technology, Hokkaido University, Sapporo, Hokkaido.
Ezo
Along with Masaya Honjo, Hiroyuki Iizuka, and Masahito Yamamoto, Kei Takada is author of the Hex playing program Ezo, which played the 17th Computer Olympiad in Yokohama 2013, and the 18th Computer Olympiad in Leiden 2015. Ezo uses 2-ply alpha-beta search with an evaluation function based on the theory of complex networks. The function combines a betweenness measure of the player’s two sides with a shortest path metric that accounts for the degree of important nodes, further elaborted by Kei Takada at the Advances in Computer Games 14 conference in Leiden 2015 [2]. EZO switches to a Hexy-style evaluation when remaining time becomes short [3]. Ezo-CNN, which played the 20th Computer Olympiad 2017, uses a convolutional neural policy network for move ordering [4] [5].
Selected Publications
- Kei Takada, Masaya Honjo, Hiroyuki Iizuka, Masahito Yamamoto (2015). Developing Computer Hex using Global and Local Evaluation based on Board Network Characteristics. Advances in Computer Games 14, slides as pdf
- Kei Takada, Masaya Honjo, Hiroyuki Iizuka, Masahito Yamamoto (2015). Developing Evaluation Function of Hex using Board Network Characteristics and SVM. Transactions of the Japanese Society for Artificial Intelligence, Vol. 30, No. 6
- Ryan Hayward, Jakub Pawlewicz, Kei Takada, Tony van der Valk (2017). MOHEX Wins 2015 Hex 11x11 and Hex 13x13 Tournaments. ICGA Journal, Vol. 39, No. 1 » 18th Computer Olympiad
- Kei Takada, Hiroyuki Iizuka, Masahito Yamamoto (2017). Reinforcement Learning for Creating Evaluation Function Using Convolutional Neural Network in Hex. TAAI 2017
- Kei Takada, Hiroyuki Iizuka, Masahito Yamamoto (2018). Computer Hex Algorithm Using a Move Evaluation Method Based on a Convolutional Neural Network. Communications in Computer and Information Science
References
- ↑ Image cropped from 18th Computer Olympid - Day 5 Photos by Jan Krabbenbos, Price giving in Hex
- ↑ Kei Takada, Masaya Honjo, Hiroyuki Iizuka, Masahito Yamamoto (2015). Developing Computer Hex using Global and Local Evaluation based on Board Network Characteristics. Advances in Computer Games 14, slides as pdf
- ↑ Ryan Hayward, Broderick Arneson, Shih-Chieh Huang, Jakub Pawlewicz (2013). MOHEX Wins Hex Tournament. ICGA Journal, Vol. 36, No. 3, pdf
- ↑ Ryan Hayward, Noah Weninger (2017). Hex 2017: MoHex wins the 11x11 and 13x13 tournaments. ICGA Journal, Vol. 39, Nos. 3-4
- ↑ Kei Takada, Hiroyuki Iizuka, Masahito Yamamoto (2018). Computer Hex Algorithm Using a Move Evaluation Method Based on a Convolutional Neural Network. Communications in Computer and Information Science
- ↑ Kei Takada's scientific contributions | Hokkaido University, Sapporo (Hokudai)
