{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/teaching-deep-convolutional-neural-networks","title":"Teaching Deep Convolutional Neural Networks to Play Go","arxiv_id":"1412.3409","date":"2014-12-10","proceeding":null,"authors":["Christopher Clark","Amos Storkey"],"abstract":"Mastering the game of Go has remained a long standing challenge to the field\nof AI. Modern computer Go systems rely on processing millions of possible\nfuture positions to play well, but intuitively a stronger and more 'humanlike'\nway to play the game would be to rely on pattern recognition abilities rather\nthen brute force computation. Following this sentiment, we train deep\nconvolutional neural networks to play Go by training them to predict the moves\nmade by expert Go players. To solve this problem we introduce a number of novel\ntechniques, including a method of tying weights in the network to 'hard code'\nsymmetries that are expect to exist in the target function, and demonstrate in\nan ablation study they considerably improve performance. Our final networks are\nable to achieve move prediction accuracies of 41.1% and 44.4% on two different\nGo datasets, surpassing previous state of the art on this task by significant\nmargins. Additionally, while previous move prediction programs have not yielded\nstrong Go playing programs, we show that the networks trained in this work\nacquired high levels of skill. Our convolutional neural networks can\nconsistently defeat the well known Go program GNU Go, indicating it is state of\nthe art among programs that do not use Monte Carlo Tree Search. It is also able\nto win some games against state of the art Go playing program Fuego while using\na fraction of the play time. This success at playing Go indicates high level\nprinciples of the game were learned.","url_abs":"http://arxiv.org/abs/1412.3409v2","url_pdf":"http://arxiv.org/pdf/1412.3409v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"teaching-deep-convolutional-neural-networks","repo_url":"https://github.com/jmgilmer/GoCNN","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"game-of-go","task_name":"Game of Go"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}