Elmo (stylized as elmo, a blend of elastic and monkey) is a computer shogi evaluation function and book file (joseki) created by Makoto Takizawa (瀧澤誠). It is designed to be used with a third-party shogi alpha–beta search engine.
Combined with the yaneura ou (やねうら王) search, Elmo became the champion of the 27th annual World Computer Shogi Championship (世界コンピュータ将棋選手権) in May 2017.[1][2] However, in the Den Ō tournament (将棋電王戦) in November 2017, Elmo was not able to make it to the top five engines losing to 平成将棋合戦ぽんぽこ (1st), shotgun (2nd), ponanza (3rd), 読み太 (4th), and Qhapaq_conflated (5th).[3]
In October 2017, DeepMind claimed that its program AlphaZero, after two hours of massively parallel training (700,000 steps or 10,300,000 games), began to exceed Elmo's performance. With a full nine hours of training (24 million games), AlphaZero defeated Elmo in a 100-game match, winning 90, losing 8, and drawing two.[4][5]
Elmo is free software that may be run on shogi engine interface GUIs such as Shogidokoro and ShogiGUI.[6][7][8]
- ^ "The 27th World Computer Shogi Championship". Computer Shogi Association. Retrieved 10 December 2017.
- ^ "第27回 世界コンピュータ将棋選手権は新星「elmo」が制覇！ ～評価関数と定跡が公開" (in Japanese). 11 May 2017. Retrieved 10 December 2017.
- ^ "第5回 将棋電王トーナメント - niconico". denou.jp.
- ^ David Silver; Thomas Hubert; Julian Schrittwieser; Ioannis Antonoglou; Matthew Lai; Arthur Guez; Marc Lanctot; Laurent Sifre; Dharshan Kumaran; Thore Graepel; Timothy Lillicrap; Karen Simonyan; Demis Hassabis (5 December 2017). "Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm". arXiv:1712.01815 [cs.AI].
- ^ "DeepMind's AI became a superhuman chess player in a few hours, just for fun". The Verge. Retrieved 2017-12-06.
- ^ "【公式】コンピュータ将棋ソフト「elmo」導入方法". mk-takizawa.github.io.
- ^ "将棋GUIソフト「将棋所」のページ". www.geocities.jp. Archived from the original on 2018-01-02. Retrieved 2017-12-10.
- ^ "ShogiGUI". shogigui.siganus.com.
© MMXXIII Rich X Search. We shall prevail. All rights reserved. Rich X Search