What is DeepEsper
DeepEsper is a computer go program developed by SkyHobbit Go club. This program is not a go game but a logic to play go. It is aimed to elucidate what is win. In addition, we are also aiming for a Go program that humans feel pleasing and enjoyable, as well as a Go program to support improvement of human skill.
We continue to expand based on programs created by participating in Computer Go training seminars held at the University of Electro-Communications in 2015. We expand the sample program of the course, so DeepEsper is influenced by the Go program "Aya".
The program source code and executable file are not opened as of 2017 now. We do not distribute or sell.
Board size 9x9 version, record with 4 wins and 2 losses in the 2015 computer go training seminar, 5th place in 16 people. It is about 1600 in rating of CGOS 9x9.
Board size 19x19 version, in the 9th UEC Cup, Rank 27th. 2 wins and 5 losses. The possibility is hidden. It is about 1000 in the rating of CGOS 19x19.
Yutaka Matsuo (2015) "Artificial intelligence is beyond human or what is in front of deep learning" Kadokawa EPUB selection book.
Hitoshi Matsubara, Kazuki Mizushi, Hiroshi Yamashita (2012) "Theory and Practice of Computer Go Monte Carlo" Kyoritsu Publishing.
Kenta Murata (2015) "AI technology of attention that learns while learning the first deep learning" "WEB + DB PRESS" Vol.89, pp. 47-77, technical commentary company.
"What A. I. Wants" "WIRED" Vol. 20, pp. 13 - 163, WIRED JAPAN.
Remi Coulom (translated by Hataki Kato) (2015), "Computing Elo Ratings of Move Patterns in the Game of Go" (translated)
History and Schedule
|2017/03/18||4.0.0||10th UEC cup entry version.|
|2016/10/02||3.0.2||2016 CGF Open19ro entry version.|
|2016/10/01||3.0.1||2016 CGF Open9ro entry version.|
|2016/03/19||2.2.5||19ro version Completion9th UEC cupentry version.|
|2015/09/12||1.0.0||9ro version Completion. Computer go Seminar mini Convention entry version.|