Chess fans and netizens for ages: Ke Jie AlphGo end world war I-Ke-Jie, AlphaGo, go-IT information
Li Shishi and AlphaGo man-machine war although it has been almost 2 months time, but the resulting AI boom is no less. Birthplace of China as Weiqi, Ke Jie as a human player first person, AlphaGo as the first master of artificial intelligence, and for the development of chess, chess fans expected and players ' feelings, Ke Jie and AlphaGo should have been in China for a game, no matter what the outcome, this is a real happy ending.
Ranked first in the world VS the next
Due to historical reasons for the development of the modern game, there is no one authoritative chess organization in the world, naturally there is no unity like chess, tennis world rankings. But after Li Shishi AlphaGo man-machine war, Europe's "Go Ratings" rankings suddenly being dug out. Standing on the point of view of a fourth party, look at the problem might be more fair.
"Go Ratings" ranking is based on national competitions and world series game the score calculated data, and is not a simple addition and subtraction, players win or lose over an opponent's score can also affect the score of his own. AlphaGo recently did not play chess, but its rank has been increasing recently. Main reason is that five-game winning streak after the man-machine war Li Shishi (as of press time had six in a row), and promoted the growth of AlphaGo score.
▲ AlphaGo points approaching Ke Jie
"Go Ratings," May 2, data AlphaGo and Ke Jie's score only 24 points. If Li Shishi to win it behind, Ke Jie occasionally losing chess, AlphaGo is likely to happen in the case does not deal with Ke Jie over Ke Jie ... ...
Go originated in China
There is a saying that goes, "modern China is the mother of go, Japan is a mother". Unfortunately is that the human Google chose the war in Korea. Of course, there are some restrictions of the laws, but to all Chinese chess player and Chess enthusiasts, there is always a bit of regret.
Weiqi is a favorite of mathematicians and physicists in the West, people seldom have to delve into chess. DeepMind when AlphaGo came, they did not think this matter in China such a sensation. Because in the European world, chess is a relatively niche game.
▲ Xiergaibulin shook hands with Li Shishi
Shake hands with Li Shishi to Google's co-founder, Sergey Brin, Stanford doctor, he is now worth more than $ 30 billion, is currently focused on the next generation of innovation in Google, explore the "moon landing" projects and ideas. Brin itself will go, or even say "compared to chess, go make me feel more philosophy of life" words.
Human wars in Korea, less than ten days ' time, there have been so many stories, filled with so much moved, set off a great go boom, telling people the arrival of a new era of science and technology. Unfortunately, Chinese netizens are just few words about some news, a lot of deep content does not come into contact with.
? Man-machine war have so many moving moments
DeepMind, their original intention might just want to do a scientific experiment. Because the AlphaGo level is too high, with their team average "no amateur 1" level, it is unable to read the AlphaGo contains detailed explanations behind the profound meaning. From the results, there is no doubt that man-machine war was a great victory for Google PR and brand marketing, go the achievements of this remarkable event.
After the man-machine war, China ushered in the development opportunities of go, but it is also a test. Many ordinary people do not understand the implications of these man-machine war, as time goes on, people's curiosity fades, Chinese chess will face what kind of situation?
So, Google at least owes the Chinese chess at a news conference. AI did not "crack" go, just human "shoutan" once, in pursuit of the game can be carried forward on the road.
Ke Jie bilishishi is also suitable for combat AlphaGo
Calculating relative winning, AlphaGo Li Shishi 4:1, Ke Jie Li Shishi was 8:2, just close.
From the technical analysis Ke Jie qifeng and Li Shishi. Li Shishi was "Wolf" way of fighting, waiting for opponent flaw, and then fight for a death, or at least in hot pursuit. Unfortunately for the AlphaGo, where it most is impeccable. Li Shishi's most powerful ability is almost useless.
▲ Ke Jie is more appropriate than the Li Shishi AlphaGo opponent
Ke Jie and Li Shishi instead, he qifeng flexibility is good at converting, give manufacturing flaws, this feature just AlphaGo can be relatively weak in the first half. Moreover, Li Shishi has helped Ke Jie had feelers put out. Ke Jie to go and challenge in March and Li Shishi AlphaGo, hope of winning greater than Li Shishi. But AlphaGo in progress all the time, Ke Jie under the earlier the better.
AlphaGo have yet to prove themselves
Man-machine World War IV, AlphaGo this mistake, to prove that he is not "perfect". Players of the ever-changing, even the collection of human intelligence, artificial intelligence miscalculations can also occur. Making mistakes is not terrible, the key is how to deal with and remedy, this is the place where people better than artificial intelligence.
Figure 1: white 1 dig Li Shishi's "hand of God", induced the AlphaGo of "miscalculation". We were attracted by the light by Li Shishi this chess, from a technology perspective, people may miss the more interesting move.
Figure 1 2:AlphaGo black "dig", is from a scientific point of view, the AlphaGo team's biggest gains. The local changes are not complex, 1 black and white 2 switched, black is not only worthless, itself also suffered serious wounds, this game is the AlphaGo the game lost lost. Ordinarily calculated in AlphaGo this error should not occur, but it is what happened, for what? Is still a mystery.
Error is not terrible, the key is to have good coping mechanisms. AlphaGo may learn to "anger management" are next to scientists to solve the problem.
It is understood that the AlphaGo against network stand-alone version of the record is about 1:3. Artificial intelligence and depth of learning technology, relies on the computing power to boost performance is not to be proud of. Future applications of AI technology is health care, automobile driving, single version of the maturity of the performance is the key to promotion. AlphaGo you need to prove your own pure "pieces" rather than relying on "computing power" accumulation level.
So AlphaGo need a new game to prove their progress.
We are ready for it?
No matter from which side, Ke Jie AlphaGo should be one war, unveiled the world's first suspense, witnessed leaps in artificial intelligence, make up for the regret of the Chinese go, but chess fans are worrying.
棋迷与网友望穿秋水：柯洁AlphGo终须一战 - 柯洁,AlphaGo,围棋 - IT资讯
李世石与AlphaGo人机大战虽然已经过了将近2个月的时间，但由此引发的人工智能热潮却丝毫不减。中国作为围棋的发源地，柯洁作为目前人类棋手的第一人，AlphaGo作为人工智能的第一高手，为了围棋的发展、棋迷的期待、棋手的情怀，柯洁和AlphaGo都应该在中国进行一次对弈，无论结果如何，这才是真正的大团圆结局。
世界排名第一VS第二
由于现代围棋发展的历史原因，目前并没有一个权威性的围棋世界组织，自然也没有像国际象棋、网球那种统一的世界排名了。不过李世石与AlphaGo人机大战之后，欧洲的“Go Ratings”排名忽然被大家挖了出来。确实站在一个第四方的角度，看待问题可能会更加公允一些。
“Go Ratings”排名是以各国的大赛和世界大赛的比赛作为等级分计算数据，而且不是简单的加减，棋手赢过或者输过对手的等级分也会影响他自己的等级分。比如AlphaGo最近没有下棋，但它的等级分最近却一直在增加。主要原因是，李世石人机大战之后五连胜(截至发稿时已经六连胜了)，带动了AlphaGo等级分的增长。
▲AlphaGo的积分渐渐逼近柯洁
据“Go Ratings”5月2日数据，AlphaGo与柯洁的等级分仅相差24分。后面如果李世石一直赢下去，而柯洁偶尔输棋的话，很有可能发生AlphaGo在不与柯洁交手的情况下，超过柯洁……
围棋起源于中国
有这样一种说法，“现代围棋中国是生母，日本是养母”。但可惜的是，人机大战谷歌却选择了在韩国进行。当然这里面有些客观条件的限制，但对所有中国棋手和围棋爱好者们来说，心里总有那么一丝遗憾。
围棋在西方是数学家和物理学家的最爱，普通人很少有去钻研围棋的。DeepMind做出AlphaGo来的时候，他们并没有想到这件事竟然在中国如此轰动。因为在欧洲人的认知里，围棋还是相对小众的游戏。
▲希尔盖布林与李世石握手
与李世石握手的是谷歌的联合创始人谢尔盖布林，斯坦福博士，他现在的身价已经超过了300亿美元，目前在谷歌专注于下一代创新，探索“登月”项目和创意。布林本身会下围棋，甚至说出了“相比国际象棋，围棋让我感受到更多人生哲理”这样话语。
人机大战在韩国，不到十天的时间，发生了那么多故事，充斥着那么多感动，掀起了巨大的围棋热潮，告诉了人们科技新时代的到来。可惜中国的网友们只是只言片语地了解到了一些消息，很多深层次的内容并没有接触到。
▲人机大战有那么多令人感动的瞬间
对DeepMind来说，他们的初衷可能只是想做一次科学实验。因为AlphaGo的围棋水平实在太高，以他们团队平均“不到业余1段”的围棋水平，实在是无法看懂AlphaGo招法背后的深刻内涵。从结果来看，毫无疑问，人机大战是谷歌公关和品牌营销的一次巨大胜利，围棋成就了这次举世瞩目的盛典。
人机大战之后，中国围棋迎来了发展的良机，但这同时也是考验。很多普通人并不了解这次人机大战背后的意义，随着时间的推移，人们对好奇心淡化，中国围棋将要面临怎样的局面呢？
所以，谷歌至少欠了中国围棋一场新闻发布会吧。人工智能并没有“破解”围棋，只是与人类“手谈”了一回，在追寻围棋极致的道路上可以相携前进。
柯洁比李世石更适合对抗AlphaGo
计算相对胜率的话，AlphaGo对李世石是4比1，柯洁对李世石是8比2，正好势均力敌。
从技术上分析李世石和柯洁的棋风。李世石是“狼”一样的战斗方式，等待对手出现破绽，然后争取一击致命，至少也是穷追不舍。可惜对AlphaGo来说，它最强的地方就是无懈可击。李世石最强大的能力，几乎没有用武之地。
▲柯洁是比李世石更适合AlphaGo的对手
柯洁则和李世石相反，他棋风灵活，擅长转换，给对手制造破绽，这样的特点恰好可以针对AlphaGo相对较弱的前半盘。而且，李世石已经帮柯洁投石问路过了。柯洁如果去挑战三月份与李世石比赛的AlphaGo，取胜的希望肯定大过李世石。不过AlphaGo每时每刻都在进步，越早下对柯洁越有利。
AlphaGo尚需证明自己
人机大战第四局，AlphaGo出现的失误，证明他并非“完美”。围棋的千变万化，即使是集人类智慧于一身的人工智能也会出现误算。犯错并不可怕，关键是如何应对和弥补，这是人类要强于人工智能的地方。
图1：白1挖是李世石的“神之一手”，诱发了AlphaGo的“误算”。大家被李世石这手棋的光芒所吸引，单从科技角度，人们可能忽略了更有趣的一步棋。
图2：AlphaGo的黑1“挖”，则是从科学角度上看，AlphaGo团队最大的收获。这个局部变化并不复杂，黑1与白2的交换，黑棋不但没有价值，还令自身受到了严重创伤，这手棋才是让AlphaGo这盘棋失利的最终败招。按说以AlphaGo的计算力不应该出现这种错误，但它就是发生了，为什么呢？至今还是个未解之谜。
出现错误并不可怕，关键是有良好的应对机制。也许让AlphaGo学会“情绪管理”是下一步科学家们要解决的问题。
据了解，AlphaGo单机版对阵网络版的战绩是1比3左右。对人工智能和深度学习技术来说，依靠计算能力来提升性能并不值得自豪。AI技术将来的应用领域在于医疗，汽车驾驶等，单机版性能的成熟才是可以推广的关键。AlphaGo下面需要证明的是自己纯粹的“棋力”，而不是依靠“计算力”累加出来的水平。
所以，AlphaGo也需要一场新的棋局来证明自己的进步。
大家都准备好了吗？
无论从哪个方面讲，柯洁与AlphaGo都应该一战，揭晓世界第一的悬念，见证人工智能的飞跃，弥补中国围棋的遗憾，了却棋迷们的牵挂。