Home  ›  Human Go Champion Backtracks on Vow to Never Face an Ai Opponent Again

Human Go Champion Backtracks on Vow to Never Face an Ai Opponent Again

Written By Saturday, March 26, 2022 Add Comment Edit

Artificial intelligence that plays Become

This article is virtually a computer program. For the moving picture, encounter AlphaGo (film).

AlphaGo logo

AlphaGo is a computer plan that plays the board game Become.[i] It was adult by DeepMind Technologies[two] a subsidiary of Google (at present Alphabet Inc.). Subsequent versions of AlphaGo became increasingly powerful, including a version that competed under the name Chief.[3] After retiring from competitive play, AlphaGo Master was succeeded by an even more than powerful version known equally AlphaGo Zero, which was completely self-taught without learning from human games. AlphaGo Zippo was then generalized into a program known every bit AlphaZero, which played additional games, including chess and shogi. AlphaZero has in turn been succeeded by a programme known as MuZero which learns without existence taught the rules.

AlphaGo and its successors utilize a Monte Carlo tree search algorithm to notice its moves based on knowledge previously acquired by machine learning, specifically by an artificial neural network (a deep learning method) by all-encompassing training, both from man and computer play.[iv] A neural network is trained to identify the all-time moves and the winning percentages of these moves. This neural network improves the strength of the tree search, resulting in stronger move option in the next iteration.

In October 2015, in a match against Fan Hui, the original AlphaGo became the first estimator Get program to trounce a man professional Go player without handicap on a full-sized 19×19 board.[v] [half dozen] In March 2016, it beat Lee Sedol in a five-game match, the first time a computer Go programme has browbeaten a 9-dan professional without handicap.[7] Although it lost to Lee Sedol in the quaternary game, Lee resigned in the final game, giving a terminal score of 4 games to 1 in favour of AlphaGo. In recognition of the victory, AlphaGo was awarded an honorary nine-dan by the Korea Baduk Clan.[8] The lead up and the claiming lucifer with Lee Sedol were documented in a documentary film also titled AlphaGo,[nine] directed past Greg Kohs. The win past AlphaGo was chosen by Science equally ane of the Breakthrough of the Yr runners-upwards on 22 December 2016.[10]

At the 2022 Future of Go Summit, the Primary version of AlphaGo beat Ke Jie, the number one ranked player in the earth at the time, in a 3-game match, after which AlphaGo was awarded professional 9-dan by the Chinese Weiqi Association.[11]

After the match between AlphaGo and Ke Jie, DeepMind retired AlphaGo, while continuing AI research in other areas.[12] The self-taught AlphaGo Zero accomplished a 100–0 victory against the early on competitive version of AlphaGo, and its successor AlphaZero is currently perceived equally the globe'south top player in Go.[13] [fourteen]

History [edit]

Go is considered much more than difficult for computers to win than other games such as chess, because its much larger branching factor makes information technology prohibitively difficult to use traditional AI methods such as blastoff–beta pruning, tree traversal and heuristic search.[v] [15]

Almost two decades afterward IBM'southward computer Deep Blue beat earth chess champion Garry Kasparov in the 1997 match, the strongest Get programs using artificial intelligence techniques only reached about apprentice 5-dan level,[four] and however could non beat out a professional person Go player without a handicap.[5] [6] [16] In 2012, the software program Zen, running on a four PC cluster, crush Masaki Takemiya (9p) twice at five- and four-rock handicaps.[17] In 2013, Crazy Stone trounce Yoshio Ishida (9p) at a four-stone handicap.[eighteen]

Co-ordinate to DeepMind's David Silverish, the AlphaGo enquiry project was formed around 2022 to test how well a neural network using deep learning tin compete at Go.[nineteen] AlphaGo represents a significant improvement over previous Get programs. In 500 games against other available Get programs, including Crazy Stone and Zen, AlphaGo running on a single computer won all but ane.[20] In a similar matchup, AlphaGo running on multiple computers won all 500 games played against other Become programs, and 77% of games played against AlphaGo running on a single reckoner. The distributed version in October 2022 was using i,202 CPUs and 176 GPUs.[iv]

Match against Fan Hui [edit]

In October 2015, the distributed version of AlphaGo defeated the European Go champion Fan Hui,[21] a 2-dan (out of 9 dan possible) professional person, 5 to aught.[6] [22] This was the first fourth dimension a computer Go plan had browbeaten a professional person human actor on a full-sized board without handicap.[23] The announcement of the news was delayed until 27 January 2022 to coincide with the publication of a paper in the journal Nature [iv] describing the algorithms used.[6]

Lucifer against Lee Sedol [edit]

AlphaGo played Due south Korean professional Go player Lee Sedol, ranked 9-dan, one of the best players at Become,[16] [ needs update ] with five games taking place at the 4 Seasons Hotel in Seoul, South korea on 9, 10, 12, 13, and 15 March 2016,[24] [25] which were video-streamed alive.[26] Out of five games, AlphaGo won four games and Lee won the fourth game which made him recorded equally the but human actor who beat out AlphaGo in all of its 74 official games.[27] AlphaGo ran on Google's cloud calculating with its servers located in the United States.[28] The match used Chinese rules with a 7.5-signal komi, and each side had two hours of thinking time plus three 60-2nd byoyomi periods.[29] The version of AlphaGo playing against Lee used a like corporeality of calculating power as was used in the Fan Hui lucifer.[30] The Economist reported that it used 1,920 CPUs and 280 GPUs.[31] At the fourth dimension of play, Lee Sedol had the second-highest number of Become international championship victories in the globe afterward South Korean thespian Lee Changho who kept the world title title for sixteen years.[32] Since there is no unmarried official method of ranking in international Go, the rankings may vary among the sources. While he was ranked top sometimes, some sources ranked Lee Sedol as the quaternary-best role player in the world at the time.[33] [34] AlphaGo was not specifically trained to face Lee nor was designed to compete with any specific human players.

The first three games were won past AlphaGo following resignations by Lee.[35] [36] However, Lee shell AlphaGo in the fourth game, winning past resignation at move 180. AlphaGo so continued to accomplish a 4th win, winning the 5th game by resignation.[37]

The prize was United states of america$one one thousand thousand. Since AlphaGo won four out of 5 and thus the series, the prize volition be donated to charities, including UNICEF.[38] Lee Sedol received $150,000 for participating in all v games and an additional $20,000 for his win in Game 4.[29]

In June 2016, at a presentation held at a university in the Netherlands, Aja Huang, i of the Deep Mind team, revealed that they had patched the logical weakness that occurred during the 4th game of the lucifer between AlphaGo and Lee, and that subsequently motion 78 (which was dubbed the "divine move" by many professionals), information technology would play equally intended and maintain Black'southward advantage. Before motion 78, AlphaGo was leading throughout the game, but Lee's move caused the program's computing powers to be diverted and confused.[39] Huang explained that AlphaGo'southward policy network of finding the most accurate motion society and continuation did not precisely guide AlphaGo to make the correct continuation after motility 78, since its value network did not decide Lee's 78th move as existence the most likely, and therefore when the move was made AlphaGo could not make the right adjustment to the logical continuation.[40]

Sixty online games [edit]

On 29 Dec 2016, a new account on the Tygem server named "Magister" (shown equally 'Magist' at the server'southward Chinese version) from South Korea began to play games with professional person players. It changed its business relationship name to "Master" on thirty Dec, then moved to the FoxGo server on 1 January 2017. On 4 January, DeepMind confirmed that the "Magister" and the "Master" were both played by an updated version of AlphaGo, called AlphaGo Master.[41] [42] As of 5 Jan 2017, AlphaGo Chief's online record was 60 wins and 0 losses,[43] including three victories over Get'south top-ranked player, Ke Jie,[44] who had been quietly briefed in advance that Master was a version of AlphaGo.[43] Afterwards losing to Principal, Gu Li offered a compensation of 100,000 yuan (US$fourteen,400) to the outset human player who could defeat Master.[42] Master played at the pace of 10 games per day. Many quickly suspected it to be an AI thespian due to petty or no resting between games. Its adversaries included many world champions such every bit Ke Jie, Park Jeong-hwan, Yuta Iyama, Tuo Jiaxi, Mi Yuting, Shi Yue, Chen Yaoye, Li Qincheng, Gu Li, Chang Hao, Tang Weixing, Fan Tingyu, Zhou Ruiyang, Jiang Weijie, Chou Chun-hsun, Kim Ji-seok, Kang Dong-yun, Park Yeong-hun, and Won Seong-jin; national champions or world championship runners-upward such every bit Lian Xiao, Tan Xiao, Meng Tailing, Dang Yifei, Huang Yunsong, Yang Dingxin, Gu Zihao, Shin Jinseo, Cho Han-seung, and An Sungjoon. All 60 games except one were fast-paced games with three 20 or xxx seconds byo-yomi. Master offered to extend the byo-yomi to one minute when playing with Nie Weiping in consideration of his age. After winning its 59th game Master revealed itself in the chatroom to be controlled by Dr. Aja Huang of the DeepMind team,[45] and so changed its nationality to the United kingdom. Later on these games were completed, the co-founder of Google DeepMind, Demis Hassabis, said in a tweet, "we're looking frontwards to playing some official, total-length games later [2017] in collaboration with Go organizations and experts".[41] [42]

Go experts were impressed past the program'due south functioning and its nonhuman play mode; Ke Jie stated that "Later on humanity spent thousands of years improving our tactics, computers tell usa that humans are completely wrong... I would go as far as to say not a single human has touched the edge of the truth of Become."[43]

Future of Become Pinnacle [edit]

In the Future of Go Summit held in Wuzhen in May 2017, AlphaGo Chief played 3 games with Ke Jie, the world No.1 ranked player, also as two games with several top Chinese professionals, one pair Go game and one against a collaborating team of five homo players.[46]

Google DeepMind offered 1.5 one thousand thousand dollar winner prizes for the 3-game lucifer between Ke Jie and Master while the losing side took 300,000 dollars.[47] [48] Master won all three games against Ke Jie,[49] [50] later on which AlphaGo was awarded professional person 9-dan by the Chinese Weiqi Association.[11]

After winning its three-game match against Ke Jie, the height-rated world Get thespian, AlphaGo retired. DeepMind also disbanded the team that worked on the game to focus on AI enquiry in other areas.[12] Afterwards the Peak, Deepmind published fifty full length AlphaGo vs AlphaGo matches, as a souvenir to the Go community.[51]

AlphaGo Cypher and AlphaZero [edit]

AlphaGo'due south team published an article in the journal Nature on 19 Oct 2017, introducing AlphaGo Zippo, a version without human data and stronger than whatsoever previous human-champion-defeating version.[52] By playing games against itself, AlphaGo Zero surpassed the strength of AlphaGo Lee in three days by winning 100 games to 0, reached the level of AlphaGo Master in 21 days, and exceeded all the former versions in forty days.[53]

In a paper released on arXiv on five December 2017, DeepMind claimed that it generalized AlphaGo Cipher's arroyo into a single AlphaZero algorithm, which achieved within 24 hours a superhuman level of play in the games of chess, shogi, and Get by defeating world-champion programs, Stockfish, Elmo, and three-day version of AlphaGo Nix in each case.[54]

Teaching tool [edit]

On xi December 2017, DeepMind released AlphaGo instruction tool on its website[55] to clarify winning rates of different Get openings as calculated past AlphaGo Master.[56] The education tool collects six,000 Go openings from 230,000 human games each analyzed with 10,000,000 simulations by AlphaGo Master. Many of the openings include human move suggestions.[56]

Versions [edit]

An early on version of AlphaGo was tested on hardware with diverse numbers of CPUs and GPUs, running in asynchronous or distributed mode. Two seconds of thinking time was given to each movement. The resulting Elo ratings are listed below.[four] In the matches with more time per motion college ratings are achieved.

Configuration and performance
Configuration Search
threads		 No. of CPU No. of GPU Elo rating
Unmarried[4] p. x–11 40 48 one two,181
Unmarried 40 48 two two,738
Single 40 48 4 2,850
Single xl 48 8 2,890
Distributed 12 428 64 2,937
Distributed 24 764 112 3,079
Distributed 40 i,202 176 3,140
Distributed 64 i,920 280 3,168

In May 2016, Google unveiled its own proprietary hardware "tensor processing units", which it stated had already been deployed in multiple internal projects at Google, including the AlphaGo friction match against Lee Sedol.[57] [58]

In the Futurity of Get Acme in May 2017, DeepMind disclosed that the version of AlphaGo used in this Height was AlphaGo Master,[59] [sixty] and revealed that it had measured the strength of different versions of the software. AlphaGo Lee, the version used against Lee, could give AlphaGo Fan, the version used in AlphaGo vs. Fan Hui, three stones, and AlphaGo Main was even three stones stronger.[61]

Configuration and forcefulness[62]
Versions Hardware Elo rating Engagement Results
AlphaGo Fan 176 GPUs,[53] distributed iii,144[52] October 2015 five:0 against Fan Hui
AlphaGo Lee 48 TPUs,[53] distributed 3,739[52] Mar 2016 4:ane against Lee Sedol
AlphaGo Master 4 TPUs,[53] single machine 4,858[52] May 2017 60:0 against professional person players;
Future of Get Summit
AlphaGo Zero (40 block) four TPUs,[53] single motorcar v,185[52] Oct 2017 100:0 against AlphaGo Lee

89:xi against AlphaGo Principal

AlphaZero (20 cake) 4 TPUs, single machine 5,018

[63]

Dec 2017 60:40 against AlphaGo Zero (20 block)

Algorithm [edit]

As of 2016, AlphaGo's algorithm uses a combination of machine learning and tree search techniques, combined with extensive training, both from human and computer play. It uses Monte Carlo tree search, guided past a "value network" and a "policy network," both implemented using deep neural network engineering.[5] [iv] A express amount of game-specific feature detection pre-processing (for example, to highlight whether a move matches a nakade pattern) is practical to the input before it is sent to the neural networks.[4] The networks are convolutional neural networks with 12 layers, trained by reinforcement learning.[64]

The system'due south neural networks were initially bootstrapped from human gameplay expertise. AlphaGo was initially trained to mimic human play past attempting to lucifer the moves of expert players from recorded historical games, using a database of effectually 30 meg moves.[21] Once information technology had reached a certain degree of proficiency, it was trained further past being set to play big numbers of games against other instances of itself, using reinforcement learning to improve its play.[five] To avoid "disrespectfully" wasting its opponent'south time, the program is specifically programmed to resign if its assessment of win probability falls beneath a certain threshold; for the match confronting Lee, the resignation threshold was gear up to 20%.[65]

Style of play [edit]

Toby Manning, the match referee for AlphaGo vs. Fan Hui, has described the program's manner as "conservative".[66] AlphaGo'southward playing style strongly favours greater probability of winning past fewer points over lesser probability of winning by more points.[19] Its strategy of maximising its probability of winning is distinct from what human players tend to do which is to maximise territorial gains, and explains some of its odd-looking moves.[67] It makes a lot of opening moves that take never or seldom been fabricated by humans. It likes to use shoulder hits, specially if the opponent is over concentrated.[ commendation needed ]

Responses to 2022 victory [edit]

[edit]

AlphaGo's March 2022 victory was a major milestone in bogus intelligence inquiry.[68] Get had previously been regarded as a hard problem in machine learning that was expected to be out of achieve for the technology of the time.[68] [69] [70] Most experts thought a Go program as powerful every bit AlphaGo was at least five years away;[71] some experts idea that it would take at least another decade before computers would trounce Go champions.[4] [72] [73] Most observers at the beginning of the 2022 matches expected Lee to beat AlphaGo.[68]

With games such every bit checkers (that has been "solved" by the Chinook draughts player team), chess, and now Go won by computers, victories at popular board games can no longer serve equally major milestones for artificial intelligence in the manner that they used to. Deep Blue's Murray Campbell chosen AlphaGo's victory "the stop of an era... board games are more or less washed and it's time to move on."[68]

When compared with Deep Blue or Watson, AlphaGo's underlying algorithms are potentially more than full general-purpose and may be testify that the scientific community is making progress towards artificial general intelligence.[nineteen] [74] Some commentators believe AlphaGo's victory makes for a proficient opportunity for social club to showtime preparing for the possible future impact of machines with general purpose intelligence. As noted by entrepreneur Guy Suter, AlphaGo but knows how to play Go and doesn't possess general-purpose intelligence; "[It] couldn't just wake up ane forenoon and determine it wants to learn how to utilize firearms."[68] AI researcher Stuart Russell said that AI systems such as AlphaGo accept progressed quicker and become more than powerful than expected, and we must therefore develop methods to ensure they "remain under human command".[75] Some scholars, such as Stephen Hawking, warned (in May 2022 before the matches) that some future self-improving AI could proceeds actual general intelligence, leading to an unexpected AI takeover; other scholars disagree: AI good Jean-Gabriel Ganascia believes that "Things similar 'common sense'... may never be reproducible",[76] and says "I don't see why we would speak about fears. On the contrary, this raises hopes in many domains such as health and infinite exploration."[75] Figurer scientist Richard Sutton said "I don't think people should be scared... but I exercise think people should be paying attention."[77]

In China, AlphaGo was a "Sputnik moment" which helped convince the Chinese government to prioritize and dramatically increase funding for artificial intelligence.[78]

In 2017, the DeepMind AlphaGo team received the inaugural IJCAI Marvin Minsky medal for Outstanding Achievements in AI. "AlphaGo is a wonderful accomplishment, and a perfect instance of what the Minsky Medal was initiated to recognise", said Professor Michael Wooldridge, Chair of the IJCAI Awards Committee. "What specially impressed IJCAI was that AlphaGo achieves what it does through a brilliant combination of classic AI techniques besides equally the state-of-the-fine art machine learning techniques that DeepMind is and then closely associated with. It's a scenic demonstration of contemporary AI, and nosotros are delighted to be able to recognise it with this award."[79]

[edit]

Become is a popular game in China, Japan and Korea, and the 2022 matches were watched by perhaps a hundred 1000000 people worldwide.[68] [80] Many top Go players characterized AlphaGo's unorthodox plays as seemingly-questionable moves that initially addled onlookers, but fabricated sense in hindsight:[72] "All but the very best Go players craft their fashion by imitating pinnacle players. AlphaGo seems to have totally original moves it creates itself."[68] AlphaGo appeared to take unexpectedly get much stronger, even when compared with its October 2022 match[81] where a reckoner had beaten a Go professional person for the outset time ever without the reward of a handicap.[82] The day subsequently Lee's first defeat, Jeong Ahram, the atomic number 82 Become contributor for one of South korea's biggest daily newspapers, said "Last night was very gloomy... Many people drank alcohol."[83] The Korea Baduk Association, the organization that oversees Become professionals in South Korea, awarded AlphaGo an honorary nine-dan title for exhibiting creative skills and pushing forward the game's progress.[84]

China's Ke Jie, an 18-year-old by and large recognized every bit the earth'due south best Go player at the time,[33] [85] initially claimed that he would exist able to beat AlphaGo, but declined to play against it for fear that information technology would "copy my style".[85] As the matches progressed, Ke Jie went back and forth, stating that "it is highly likely that I (could) lose" after analysing the first three matches,[86] just regaining conviction after AlphaGo displayed flaws in the fourth match.[87]

Toby Manning, the referee of AlphaGo's match confronting Fan Hui, and Hajin Lee, secretary full general of the International Go Federation, both reason that in the future, Go players will get help from computers to larn what they have done wrong in games and improve their skills.[82]

Subsequently game 2, Lee said he felt "speechless": "From the very beginning of the match, I could never manage an upper hand for one single motion. It was AlphaGo's full victory."[88] Lee apologized for his losses, stating after game three that "I misjudged the capabilities of AlphaGo and felt powerless."[68] He emphasized that the defeat was "Lee Se-dol's defeat" and "not a defeat of mankind".[27] [76] Lee said his eventual loss to a machine was "inevitable" only stated that "robots will never understand the beauty of the game the same manner that nosotros humans do."[76] Lee called his game four victory a "priceless win that I (would) not exchange for anything."[27]

Like systems [edit]

Facebook has too been working on its own Go-playing arrangement darkforest, also based on combining automobile learning and Monte Carlo tree search.[66] [89] Although a strong player against other calculator Go programs, every bit of early on 2016, it had not nevertheless defeated a professional man role player.[90] Darkforest has lost to CrazyStone and Zen and is estimated to be of similar strength to CrazyStone and Zen.[91]

DeepZenGo, a organization developed with back up from video-sharing website Dwango and the University of Tokyo, lost two–ane in Nov 2022 to Go master Cho Chikun, who holds the record for the largest number of Become title wins in Japan.[92] [93]

A 2022 paper in Nature cited AlphaGo'south approach as the basis for a new means of calculating potential pharmaceutical drug molecules.[94] [95]

Example game [edit]

AlphaGo Master (white) v. Tang Weixing (31 December 2016), AlphaGo won by resignation. White 36 was widely praised.

Impacts on Become [edit]

The documentary flick AlphaGo [nine] [96] raised hopes that Lee Sedol and Fan Hui would have benefitted from their experience of playing AlphaGo, merely every bit of May 2022 their ratings were lilliputian changed; Lee Sedol was ranked 11th in the world, and Fan Hui 545th.[97] On 19 November 2019, Lee appear his retirement from professional play, arguing that he could never be the top overall role player of Go due to the increasing dominance of AI. Lee referred to them as being "an entity that cannot be defeated".[98]

See as well [edit]

  • Chinook (draughts actor), draughts playing program
  • Glossary of artificial intelligence
  • Become and mathematics
  • Leela (software)
  • TD-Gammon, backgammon neural network
  • Pluribus (poker bot)
  • AlphaZero
  • AlphaFold

References [edit]

  1. ^ "Artificial intelligence: Google'south AlphaGo beats Go chief Lee Se-dol". BBC News. 12 March 2016. Retrieved 17 March 2016.
  2. ^ "DeepMind AlphaGO". DeepMind Bogus Intelligence AlphaGo.
  3. ^ "AlphaGo | DeepMind". DeepMind.
  4. ^ a b c d e f g h i Silver, David; Huang, Aja; Maddison, Chris J.; Guez, Arthur; Sifre, Laurent; Driessche, George van den; Schrittwieser, Julian; Antonoglou, Ioannis; Panneershelvam, Veda; Lanctot, Marc; Dieleman, Sander; Grewe, Dominik; Nham, John; Kalchbrenner, Nal; Sutskever, Ilya; Lillicrap, Timothy; Leach, Madeleine; Kavukcuoglu, Koray; Graepel, Thore; Hassabis, Demis (28 January 2016). "Mastering the game of Get with deep neural networks and tree search". Nature. 529 (7587): 484–489. Bibcode:2016Natur.529..484S. doi:x.1038/nature16961. ISSN 0028-0836. PMID 26819042. S2CID 515925. closed access
  5. ^ a b c d due east "Enquiry Blog: AlphaGo: Mastering the ancient game of Get with Auto Learning". Google Research Weblog. 27 January 2016.
  6. ^ a b c d "Google achieves AI 'breakthrough' by chirapsia Go champion". BBC News. 27 Jan 2016.
  7. ^ "Match one – Google DeepMind Challenge Match: Lee Sedol vs AlphaGo". 8 March 2016.
  8. ^ "Google'due south AlphaGo gets 'divine' Go ranking". The Straits Times. straitstimes.com. 15 March 2016. Retrieved 9 December 2017.
  9. ^ a b "AlphaGo Movie". AlphaGo Movie.
  10. ^ "From AI to protein folding: Our Breakthrough runners-up". Scientific discipline. 22 December 2016. Retrieved 29 Dec 2016.
  11. ^ a b "中国围棋协会授予AlphaGo职业九段 并颁发证书" (in Chinese). Sohu.com. 27 May 2017. Retrieved 9 Dec 2017.
  12. ^ a b Metz, Cade (27 May 2017). "After Win in China, AlphaGo'southward Designers Explore New AI". Wired.
  13. ^ "AlphaZero Crushes Stockfish In New one,000-Game Friction match". 17 April 2019.
  14. ^ "A full general reinforcement learning algorithm that masters chess, shogi, and Get through self-play". 7 December 2018.
  15. ^ Schraudolph, Nicol N.; Terrence, Peter Dayan; Sejnowski, J., Temporal Deviation Learning of Position Evaluation in the Game of Go (PDF)
  16. ^ a b "Estimator scores big win against humans in ancient game of Go". CNN. 28 January 2016. Retrieved 28 January 2016.
  17. ^ "Zen calculator Go program beats Takemiya Masaki with just 4 stones!". Become Game Guru. Archived from the original on 1 February 2016. Retrieved 28 January 2016.
  18. ^ "「アマ六段の力。天才かも」囲碁棋士、コンピューターに敗れる 初の公式戦". MSN Sankei News. Archived from the original on 24 March 2013. Retrieved 27 March 2013.
  19. ^ a b c John Riberio (14 March 2016). "AlphaGo's unusual moves prove its AI prowess, experts say". PC Earth . Retrieved eighteen March 2016.
  20. ^ "Google AlphaGo AI clean sweeps European Become champion". ZDNet. 28 January 2016. Retrieved 28 January 2016.
  21. ^ a b Metz, Cade (27 January 2016). "In Major AI Breakthrough, Google Organisation Secretly Beats Height Thespian at the Ancient Game of Become". WIRED . Retrieved 1 February 2016.
  22. ^ "Special Computer Go insert covering the AlphaGo v Fan Hui friction match" (PDF). British Go Journal. 2017. Retrieved ane Feb 2016.
  23. ^ "Première défaite d'un professionnel du become contre une intelligence artificielle". Le Monde (in French). 27 January 2016.
  24. ^ "Google's AI AlphaGo to have on world No ane Lee Sedol in alive circulate". The Guardian. five February 2016. Retrieved fifteen February 2016.
  25. ^ "Google DeepMind is going to take on the world's best Get player in a luxury 5-star hotel in Republic of korea". Business Insider. 22 February 2016. Retrieved 23 February 2016.
  26. ^ Novet, Jordan (4 February 2016). "YouTube will livestream Google'southward AI playing Get superstar Lee Sedol in March". VentureBeat . Retrieved 7 February 2016.
  27. ^ a b c Yoon Sung-won (14 March 2016). "Lee Se-dol shows AlphaGo beatable". The Korea Times . Retrieved xv March 2016.
  28. ^ "李世乭：即使Alpha Go得到升级也一样能赢". JoongAng Ilbo (in Chinese). 23 Feb 2016. Retrieved 24 Feb 2016.
  29. ^ a b "이세돌 vs 알파고, '구글 딥마인드 챌린지 매치' 기자회견 열려" (in Korean). Korea Baduk Association. 22 Feb 2016. Archived from the original on 3 March 2016. Retrieved 22 February 2016.
  30. ^ Demis Hassabis [@demishassabis] (11 March 2016). "We are using roughly same amount of compute power as in Fan Hui match: distributing search over farther machines has diminishing returns" (Tweet). Retrieved 14 March 2016 – via Twitter.
  31. ^ "Showdown". The Economist . Retrieved nineteen November 2016.
  32. ^ Steven Borowiec (9 March 2016). "Google'due south AI machine v world champion of 'Get': everything you need to know". The Guardian . Retrieved 15 March 2016.
  33. ^ a b Rémi Coulom. "Rating List of 2016-01-01". Archived from the original on 18 March 2016. Retrieved 18 March 2016.
  34. ^ "Korean Become master proves human intuition still powerful in Get". The Korean Herald/ANN. xiv March 2016. Archived from the original on 12 April 2016. Retrieved 15 March 2016.
  35. ^ "Google'southward AI beats world Become champion in first of 5 matches – BBC News". BBC Online . Retrieved 9 March 2016.
  36. ^ "Google AI wins 2nd Go game against globe champion – BBC News". BBC Online . Retrieved 10 March 2016.
  37. ^ "Google DeepMind AI wins final Become match for 4–ane series win". Engadget . Retrieved 15 March 2016.
  38. ^ "Human champion certain he'll beat out AI at ancient Chinese game". Associated Press. 22 February 2016. Retrieved 22 Feb 2016.
  39. ^ "In Two Moves, AlphaGo and Lee Sedol Redefined the Futurity". WIRED . Retrieved 12 November 2017.
  40. ^ "黄士杰：AlphaGo李世石人机大战第四局问题已解决date=8 July 2016" (in Chinese). Retrieved 8 July 2016.
  41. ^ a b Demis Hassabis (4 January 2017). "Demis Hassabis on Twitter: "Excited to share an update on #AlphaGo!"". Demis Hassabis's Twitter account. Retrieved 4 January 2017.
  42. ^ a b c Elizabeth Gibney (four January 2017). "Google reveals secret exam of AI bot to beat top Become players". Nature. 541 (7636): 142. Bibcode:2017Natur.541..142G. doi:10.1038/nature.2017.21253. PMID 28079098.
  43. ^ a b c "Humans Mourn Loss Afterwards Google Is Unmasked as China's Become Master". Wall Street Journal. five January 2017. Retrieved 6 January 2017.
  44. ^ "The world'south all-time Get histrion says he still has "one last motion" to defeat Google's AlphaGo AI". Quartz. four January 2017. Retrieved half dozen Jan 2017.
  45. ^ "横扫中日韩棋手斩获59胜的Master发话：我是阿尔法狗" (in Chinese). 澎湃新闻. 4 January 2017. Retrieved 11 Dec 2017.
  46. ^ "Exploring the mysteries of Get with AlphaGo and People's republic of china'southward pinnacle players". 10 Apr 2017.
  47. ^ "Earth No.1 Get player Ke Jie takes on upgraded AlphaGo in May". 10 Apr 2017.
  48. ^ "Ke Jie vs. AlphaGo: 8 things you must know". 27 May 2017.
  49. ^ Metz, Cade (23 May 2017). "Revamped AlphaGo Wins Showtime Game Against Chinese Get Grandmaster". Wired.
  50. ^ Metz, Cade (25 May 2017). "Google'southward AlphaGo Continues Say-so With Second Win in China". Wired.
  51. ^ "Full length games for Go players to savor". Deepmind. Retrieved 28 May 2017.
  52. ^ a b c d e Silver, David; Schrittwieser, Julian; Simonyan, Karen; Antonoglou, Ioannis; Huang, Aja; Guez, Arthur; Hubert, Thomas; Bakery, Lucas; Lai, Matthew; Bolton, Adrian; Chen, Yutian; Lillicrap, Timothy; Fan, Hui; Sifre, Laurent; Driessche, George van den; Graepel, Thore; Hassabis, Demis (19 October 2017). "Mastering the game of Go without human knowledge" (PDF). Nature. 550 (7676): 354–359. Bibcode:2017Natur.550..354S. doi:10.1038/nature24270. ISSN 0028-0836. PMID 29052630. S2CID 205261034. closed access
  53. ^ a b c d due east "AlphaGo Zero: Learning from scratch". DeepMind official website. 18 October 2017. Retrieved 19 October 2017.
  54. ^ Silvery, David; Hubert, Thomas; Schrittwieser, Julian; Antonoglou, Ioannis; Lai, Matthew; Guez, Arthur; Lanctot, Marc; Sifre, Laurent; Kumaran, Dharshan; Graepel, Thore; Lillicrap, Timothy; Simonyan, Karen; Hassabis, Demis (5 Dec 2017). "Mastering Chess and Shogi by Cocky-Play with a General Reinforcement Learning Algorithm". arXiv:1712.01815 [cs.AI].
  55. ^ "AlphaGo teaching tool". DeepMind.
  56. ^ a b "AlphaGo教学工具上线 樊麾：使用Master版本" (in Chinese). Sina.com.cn. 11 December 2017. Retrieved 11 December 2017.
  57. ^ McMillan, Robert (xviii May 2016). "Google Isn't Playing Games With New Flake". The Wall Street Journal . Retrieved 26 June 2016.
  58. ^ Jouppi, Norm (18 May 2016). "Google supercharges automobile learning tasks with TPU custom fleck". Google Cloud Platform Blog . Retrieved 26 June 2016.
  59. ^ "AlphaGo官方解读让三子 对人类高手没这种优势" (in Chinese). Sina. 25 May 2017. Retrieved two June 2017.
  60. ^ "各版alphago实力对比 master能让李世石版3子" (in Chinese). Sina. 24 May 2017. Retrieved ii June 2017.
  61. ^ "New version of AlphaGo cocky-trained and much more efficient". American Go Association. 24 May 2017. Retrieved one June 2017.
  62. ^ "【柯洁战败解密】AlphaGo Master最新架构和算法，谷歌云与TPU拆解" (in Chinese). Sohu. 24 May 2017. Retrieved i June 2017.
  63. ^ Silver, David; Hubert, Thomas; Schrittwieser, Julian; Antonoglou, Ioannis; Lai, Matthew; Guez, Arthur; Lanctot, Marc; Sifre, Laurent; Kumaran, Dharshan; Graepel, Thore; Lillicrap, Timothy; Simonyan, Karen; Hassabis, Demis (7 December 2018). "A full general reinforcement learning algorithm that masters chess, shogi, and Become through cocky-play". Science. 362 (6419): 1140–1144. Bibcode:2018Sci...362.1140S. doi:10.1126/science.aar6404. PMID 30523106. S2CID 54457125 – via science.org (Atypon).
  64. ^ Silver, David; Schrittwieser, Julian; Simonyan, Karen; Antonoglou, Ioannis; Huang, Aja; Guez, Arthur; Hubert, Thomas; Bakery, Lucas; Lai, Matthew; Bolton, Adrian; Chen, Yutian; Lillicrap, Timothy; Fan, Hui; Sifre, Laurent; Driessche, George van den; Graepel, Thore; Hassabis, Demis (19 October 2017). "Mastering the game of Get without human knowledge" (PDF). Nature. 550 (7676): 354–359. Bibcode:2017Natur.550..354S. doi:10.1038/nature24270. ISSN 0028-0836. PMID 29052630. S2CID 205261034. AlphaGo Lee... 12 convolutional layers
  65. ^ Cade Metz (thirteen March 2016). "Go Grandmaster Lee Sedol Grabs Consolation Win Against Google'southward AI". Wired News . Retrieved 29 March 2016.
  66. ^ a b Gibney, Elizabeth (27 January 2016). "Google AI algorithm masters ancient game of Go". Nature. 529 (7587): 445–6. Bibcode:2016Natur.529..445G. doi:ten.1038/529445a. PMID 26819021.
  67. ^ Chouard, Tanguy (12 March 2016). "The Go Files: AI computer clinches victory against Become champion". Nature. doi:ten.1038/nature.2016.19553. S2CID 155164502.
  68. ^ a b c d eastward f m h Steven Borowiec; Tracey Lien (12 March 2016). "AlphaGo beats man Go champ in milestone for artificial intelligence". Los Angeles Times . Retrieved 13 March 2016.
  69. ^ Connor, Steve (27 January 2016). "A estimator has beaten a professional at the world's most complex board game". The Independent . Retrieved 28 January 2016.
  70. ^ "Google'due south AI beats human champion at Get". CBC News. 27 January 2016. Retrieved 28 January 2016.
  71. ^ Dave Gershgorn (12 March 2016). "GOOGLE'Due south ALPHAGO BEATS WORLD CHAMPION IN Third MATCH TO WIN Entire Serial". Popular Science . Retrieved 13 March 2016.
  72. ^ a b "Google DeepMind computer AlphaGo sweeps human champ in Go matches". CBC News. Associated Press. 12 March 2016. Retrieved 13 March 2016.
  73. ^ Sofia Yan (12 March 2016). "A Google estimator victorious over the world's 'Go' champion". CNN Money . Retrieved thirteen March 2016.
  74. ^ "AlphaGo: Google's artificial intelligence to have on globe champion of aboriginal Chinese lath game". Australian Dissemination Corporation. 8 March 2016. Retrieved xiii March 2016.
  75. ^ a b Mariëtte Le Roux (12 March 2016). "Rising of the Machines: Keep an centre on AI, experts warn". Phys.org . Retrieved 13 March 2016.
  76. ^ a b c Mariëtte Le Roux; Pascale Mollard (viii March 2016). "Game over? New AI challenge to human being smarts (Update)". phys.org . Retrieved 13 March 2016.
  77. ^ Tanya Lewis (11 March 2016). "An AI expert says Google's Go-playing program is missing 1 key feature of human intelligence". Business Insider . Retrieved thirteen March 2016.
  78. ^ Mozur, Paul (xx July 2017). "Beijing Wants A.I. to Be Made in Prc by 2030". The New York Times . Retrieved eleven April 2018.
  79. ^ "Marvin Minsky Medal for Outstanding Achievements in AI". International Joint Conference on Artificial Intelligence. 19 October 2017. Retrieved 21 October 2017.
  80. ^ CHOE SANG-HUN (16 March 2016). "Google's Calculator Program Beats Lee Se-dol in Go Tournament". The New York Times . Retrieved 18 March 2016. More than 100 1000000 people watched the AlphaGo-Lee matches, Mr. Hassabis said.
  81. ^ John Ribeiro (12 March 2016). "Google's AlphaGo AI programme strong but not perfect, says defeated South Korean Go thespian". PC Globe . Retrieved 13 March 2016.
  82. ^ a b Gibney, Elizabeth (2016). "Go players react to computer defeat". Nature. doi:ten.1038/nature.2016.19255. S2CID 146868978.
  83. ^ Zastrow, Mark (xv March 2016). "How victory for Google's Become AI is stoking fear in South korea". New Scientist . Retrieved xviii March 2016.
  84. ^ JEE HEUN KAHNG; SE YOUNG LEE (15 March 2016). "Google artificial intelligence program beats Due south. Korean Go pro with 4–1 score". Reuters . Retrieved 18 March 2016.
  85. ^ a b Neil Connor (11 March 2016). "Google AlphaGo 'can't beat me' says Communist china Get grandmaster". The Telegraph (UK) . Retrieved 13 March 2016.
  86. ^ "Chinese Go master Ke Jie says he could lose to AlphaGo : The DONG-A ILBO". Retrieved 17 March 2016.
  87. ^ "...if today'south performance was its true capability, and so it doesn't deserve to play against me". One thousand.hankooki.com. 14 March 2016. Retrieved five June 2018.
  88. ^ CHOE SANG-HUN (15 March 2016). "In Seoul, Become Games Spark Interest (and Business organization) About Artificial Intelligence". The New York Times . Retrieved 18 March 2016.
  89. ^ Tian, Yuandong; Zhu, Yan (2015). "Better Calculator Go Player with Neural Network and Long-term Prediction". arXiv:1511.06410v1 [cs.LG].
  90. ^ HAL 90210 (28 Jan 2016). "No Go: Facebook fails to spoil Google's big AI day". The Guardian. ISSN 0261-3077. Retrieved 1 February 2016.
  91. ^ "Strachey Lecture – Dr Demis Hassabis". The New Livestream . Retrieved 17 March 2016.
  92. ^ "Get master Cho wins best-of-three series against Nippon-made AI". The Japan Times Online. 24 Nov 2016. Retrieved 27 November 2016.
  93. ^ "Humans strike back: Korean Get master bests AI in board game bout". CNET . Retrieved 27 November 2016.
  94. ^ "Get and make some drugs The Engineer". www.theengineer.co.united kingdom. 3 Apr 2018. Retrieved 3 April 2018.
  95. ^ Segler, Martwin H.South.; Preuss, Mike; Waller, Mark P. (29 March 2018). "Planning chemical syntheses with deep neural networks and symbolic AI". Nature. 555: 604–610.
  96. ^ "AlphaGo (2017)". Rotten Tomatoes . Retrieved 5 June 2018.
  97. ^ "Become Ratings". Go Ratings. Retrieved 5 June 2018.
  98. ^ Vincent, James (27 November 2019). "Former Go champion browbeaten past DeepMind retires after declaring AI invincible". The Verge . Retrieved 28 November 2019.

External links [edit]

bergeronyoushothe.blogspot.com

Source: https://en.wikipedia.org/wiki/AlphaGo

0 Response to "Human Go Champion Backtracks on Vow to Never Face an Ai Opponent Again"

Post a Comment

Subscribe to: Post Comments (Atom)

Iklan Atas Artikel

Iklan Tengah Artikel 1

Iklan Tengah Artikel 2

Iklan Bawah Artikel