Zhou Junxun local computer stereotypes do not lose cautious about Li Shishi Duncancoscoqd

Zhou Junxun: local computer stereotypes do not lose Duncan cautiously optimistic about the Chinese Taipei "Li Shishi red face chess Zhou Junxun silently watching the women’s game Cleveland in February 28th Jiangsu Huaian electric (reporter Deng Yu   2016) in the international intellectual movement alliance intelligence games games, the Taipei red face chess Zhou Junxun nine said in an interview, he once firmly optimistic about Li Shishi 5 to 0 victory over Google Alpha Go game software, but recently read some reports especially some analysis of science and technology circles, his heart had a shake, cautious optimistic about Li Shishi win the" Gu Li war". According to the beginning of this year a Qijie news, in October 2015 Google Alpha Go software two to 5 go to 0 victory over the European Championship Chess fan Hui, this and the previous top chess software only and amateur master blows, unable to challenge the occupation players formed a huge contrast. Zhou Junxun believes that through the Google software to win 5 game of chess fan’s point of view, under certain conditions, especially the end game stage, the computer accuracy improved significantly, "is not an exaggeration to say that each local setting is not any better than the heyday of the promotion of" stone "Li Changgao (South Korean chess representatives, once in the end game magic run amuck everywhere. Won the 17 world championships); if this progress continues, the fighting will significantly improve." According to him, the relevant foreign reports have disclosed Google go software is now using two supercomputers to simulate human combat, every day on the amount of hundreds of discs, each chess computer cost is about 300 dollars, "this investment is big companies like Google dare investment. Some people said that the 5 sets of games did not appear complex killing and looting, leaving a lot of suspense. But after the evolution of computer software, I don’t think it will make mistakes in this regard." He believes that compared to the Google Alpha Li Shishi strengths Go chess software is the biggest human creativity and imagination, set some very "rather baffling" tactics is not necessarily a bad game, will be of value evaluation system of Alpha software Go go forward real challenge. According to this view, the Google Corporation providing a prize of $1 million compared to the previous cost, only a small investment, but also through the top sample Gu Li II "can help Alpha go Go software provides verification calculation depth, testing the value evaluation system. "If you were swept away by Li Shishi 5 to 0, how long would you think the computer software would pose a substantial challenge to the top players in the profession?" In the face of this problem, Zhou Junxun tend to be more optimistic about the progress of computer software, "estimated by two years, in the beginning of next year, occupation top players may be in big trouble." 周俊勋:电脑局部定型不输石佛 谨慎看好李世石 中华台北“红面棋王”周俊勋默默关注女子比赛   华奥星空2月28日江苏淮安电(记者  邓宇)在2016国际智力运动联盟智力运动精英赛比赛间隙,中华台北“红面棋王”周俊勋九段接受采访时表示,他曾经坚定不移地看好李世石以5比0完胜谷歌Alpha Go围棋软件,但最近看了一些报道尤其是科技界人士的一些分析后,他内心有过一丝动摇,谨慎看好李世石赢得这场“谷李大战”。   据今年年初轰动棋界的新闻,在2015年10月谷歌Alpha Go围棋软件就以5比0完胜欧洲围棋冠军樊麾二段,这和以前最顶尖的围棋软件只能和业余高手过招、无法挑战职业棋手形成巨大反差。周俊勋认为,通过谷歌软件战胜樊麾的5局棋谱来看,在特定条件下尤其是官子阶段,计算机的精准度明显提升,“不夸张地说,每个局部的定型收束不见得比全盛时代的‘石佛’李昌镐(韩国围棋代表人物,曾经以官子神功横行天下,获得过17个世界冠军)差;如果再这么进步下去,中盘战斗力也会显著提升。”   据他介绍,有关国外报道曾披露谷歌围棋软件目前正用两台超级计算机模拟人类实战,每天对局量上百盘,每盘棋的计算机投入成本大约是300美金,“这样的投入也就是谷歌这样的大公司敢投入。之前有人说那5盘对局没出现过复杂对杀和打劫,留下不少悬念。但计算机软件在不断进化之后,我不认为它会在这方面犯错。”他认为李世石相对于谷歌Alpha Go围棋软件最大的长处就是人类独有的创造力和想象力,中盘时有些很“莫名其妙”的招法不一定是坏棋,会对Alpha Go围棋软件的价值评估体系提出真正挑战。照这样的说法,目前谷歌公司提供的100万美元的奖金相对于前期投入成本来说,只是小投入,而且通过“谷李大战”能帮Alpha Go围棋软件提供验证计算深度、检验价值评估体系的顶尖样本。   “如果这次被李世石5比0横扫,你觉得计算机软件多长时间会对职业顶尖棋手构成实质性挑战?”面对这个问题,周俊勋更倾向于看好计算机软件的进步速度,“估计用不了两年,就在明年年初吧,职业顶尖棋手可能就会遇到大麻烦了。”相关的主题文章: