解读AlphaGo背后的人工智能技术

Interpretation of the arti?cial intelligence technology behind Alphago

中文关键词  AlphaGo  深度学习  价值网络  策略网络

英文关键词  AlphaGo  deep learning  value network  policy network

中文摘要

      随着人工智能在各个领域的应用,越来越多的问题通过人工智能得到更优的解决, 但是围棋因其本身的复 杂度一直是人工智能领域的难解之题. AlphaGo团队利用了人工智能中的一个重要分支—深度学习训练了一款围 棋人工智能程序, 并在2016年3月与职业九段选手李世石的对弈中以4:1的比分获胜, 受到了大众的广泛关注. 本文 介绍了AlphaGo这一程序背后的复杂的网络构造以及不同网络的优缺点.

英文摘要

      With the application of arti?cial intelligence in various ?elds, more and more problems have been solved. But computer Go has been a dif?cult problem in the ?eld of arti?cial intelligence, because of the complexity of the game. AlphaGo team has trained a Go AI program which took advantage of an important branch of arti?cial intelligence – deep learning. In March 2016 AlphaGo won 4–1 the game with professional Go player Lee se-dol (9P), received extensive attention of the public.