[Introduction to Zhiyuan Xin] Just now, Dr. DeepMind Huang Shijie announced that the AlphaGo project has officially ended. If someone must put an end to AlphaGo’s research, Huang Shijie deserves it. Huang Shijie was the best spokesperson for AlphaGo when it first appeared in the public eye. He raised his hand, made a move, acted with the intelligence of a machine, and walked out of a generation of masters. He is calm, he is calm, and he is the "human arm" of AlphaGo. His departure and curtain call represent the end of an era and a more exciting beginning.
On the coldest day of the winter in December, this farewell speech in the middle of the night set off a circle of friends, and the lively and noisy discussion reflected the loneliness and sadness at the end of an era. Huang Shijie, an AI engineer who is proficient in Go, has achieved the pinnacle of AlphaGo, and his departure may also give a glimpse of the future of mankind. A group of legends in the great era made a curtain call. He was lonely, and he was replaced by his own AI. If someone must put an end to the research of AlphaGo, Huang Shijie deserved it. Huang Shijie has been the best spokesperson for AlphaGo since it first appeared in the public eye. He raised his hand, made a move, acted with the intelligence of a machine, and walked out of a generation of masters. He is calm, he is calm, and he is the "human arm" of AlphaGo. The advancing pace of the times never stops, and the flourishing age of artificial intelligence is coming. Huang Shijie, the "human arm", has changed his post and bid farewell to AlphaGo affectionately, all of which have left a splendid stroke on the milestone of AI Go.
With regard to future development, Huang Shijie said that besides Go, DeepMind is interested in medical care and robotics, hoping to make the world a better place. As for his own personal next step, of course, he will choose an interested field to invest in. Some media at the scene once again asked whether the next goal was to challenge the simulated strategy game fields of StarCraft and League of Legends. Huang Shijie did not say anything, saying, "It is still under discussion".
On November 27th, Huang Shijie forwarded a message on Facebook that DeepMind will join some top research institutions in AI research to diagnose breast cancer. DeepMind will use the latest machine learning technology, hoping to detect the signs of cancer quickly and accurately, and help doctors find cancer as soon as possible, so as to treat it effectively earlier. This may catch the wind and realize the dynamic trend after Huang Shijie.
Huang Shijie, he is AlphaGo himself.
Huang Shijie was born in Taiwan Province. In 1997, he was admitted to national chiao tung university, Taiwan Province for undergraduate study, majoring in computer and information science. From 2001 to 2003, he studied computer science and information engineering at National Taiwan Province Normal University. From 2004 to 2011, I completed my doctoral studies in National Taiwan Province Normal University, and my doctoral thesis was "A New Heuristic Algorithm for Monte Carlo Tree Search in Computer Go".
Huang Shijie specializes in the field of artificial intelligence, and all his doctoral and master’s theses were completed under the guidance of Lin Shunxi, a professor at National Taiwan Normal University. According to his tutor Lin Shunxi, Huang Shijie was a very hard-working student and was very interested in Weiqi. He was the founding president of Weiqi Club of National Taiwan Normal University, and he was qualified as an amateur 6-stage Weiqi player when he was studying.
Lin Shunxi revealed that Huang Shijie could have finished his Ph.D. in five years, but in order to continue his research and development achievements, he studied for two more years. In the seventh grade of his doctoral class, the Go program "Erica" he developed beat the internationally recognized strongest program "Zen" at that time, and won the 19-way computer Go gold medal in the international computer Olympia held in Japan.
After graduation, Dr. Huang Shijie worked as a researcher at the University of Alberta for one year, and then joined DeepMind as a senior researcher in 2012. On January 26th, 2014, Google bought the company that will shine in the future for $500 million. Since then, Huang Shijie has worked as a research scientist at Google DeepMind. In February of the same year, the AlphaGo project was officially launched, and the team consisted of only three people: Hassabis, Silva and Huang Shijie.
At the beginning of the project, the team decided to explore along the direction of deep learning and reinforcement learning. Undoubtedly, this road will be extremely difficult. However, they still hit the road.
In June, 2014, AlphaGo learned a lot of human chess manuals through convolutional neural networks, and could make a chess intuition comparable to that of human beings within 3 milliseconds. Huang Shijie named it "Strategy Network" and made continuous training and optimization on GPU.
A year later, AlphaGo’s ability to read chess has improved a lot, and it can retrieve various changes in the situation, find the best way to deal with them, and calculate the next 40-60 moves.
Since then, more and more deep learning experts have joined this team with only three people at the beginning. In August, AlphaGo will master the most critical ability to dominate the world of Go: situation judgment. AlphaGo trains the value network through self-playing, and then can judge the winning rate behind each chess game. This forms the ability to judge the situation. According to Huang Shijie’s memory, he already felt the power of AlphaGo. In October, Fan Hui fought AlphaGo, and it was Huang Shijie who replaced AlphaGo. Fan Hui lost. But this news was not announced at the time.
On January 27th, 2016, Nature magazine introduced AlphaGo of DeepMind team and the news that Fan Hui lost. Huang Shijie and Silva are the first authors of this paper.
Two months later, AlphaGo defeated Li Shizhen. This time, Huang Shijie still replaced AlphaGo. This is also the first time that Huang Shijie appeared in front of the audience.
At the end of this year, a chess player named Master killed all the Go masters in succession. After winning 59 consecutive games, Master announced his true identity: "I am Dr. Huang from AlphaGo". Huang Shijie, Huang Shijie again.
On May 23, 2017, Huang Shijie was still sitting opposite Ke Jie. After the war, Ke Jie said, "Dr. Huang’s coming to play chess may make me feel more at war with AI. I especially admire Dr. Huang. Everyone can see that I am very active when watching the live broadcast. I always like to exercise, and Dr. Huang always does not move, does not go to the toilet, does not drink water, and does not eat … So when playing chess with Dr. Huang, I think Dr. Huang is AlphaGo. "
AlphaGo fought all the way, and Huang Shijie was always on the side. It can even be said that he was AlphaGo itself.
Biography of AlphaGo
Name: AlphaGo(Fan, Lee, Master, Zero)
Alias: Teacher A, Alpha Dog
Birthday: 2014
Place of birth: London, England
1. Defeat Fan Hui
In October 2015, AlphaGo beat Fan Hui, becoming the first computer Go program that can beat professional Go players on the 19th board without giving way, and wrote down the history. Related achievements were published in Nature in January 2016.
2. Beat Li Shishi
In March, 2016, AlphaGo beat Li Shishi, a top professional chess player, 4: 1 in a five-chess match, becoming the first computer Go program to beat a professional nine-stage chess player without the help of a ceder, making history again. After five games, the Korean Chess Institute awarded AlphaGo the first honorary career in history.
3. The ranking briefly surpassed Ke Jie.
On July 18th, 2016, AlphaGo ranked first in the world on Go Ratings website. But it was overtaken by Ke Jie a few days later.
4, Alpha Go alias "Master” 60 consecutive victories
On the penultimate day of 2016, a Go user named "Master" continued to kill on the Go platform, including Ke Jie, Park Junghwan, Yuta Iyama, Chen Yaoye, Shen Zhenluan, Chang Hao, Gu Li and other Go masters, and finally ended in a 60-game winning streak. Ke Jie lost three games in a row, during which he was hospitalized for acute gastroenteritis …
In the 59th scene, "Master" introduced himself: "I’m Dr. Huang from AlphaGo".
5. AlphaGo defeated Ke Jie, and Ke Jie sobbed.
On May 27, 2017, at the Wuzhen Go Summit, Ke Jie lost to AlphaGo with a total score of 0: 3. During the competition, Ke Jie left for a long time at an unfavorable situation, and then came back to the scene in tears. After the game, Ke Jie once choked and said, It’s perfect. I’m in pain and I can’t see any hope of victory.
6, AlphaGo Zero turned out to be a self-taught Master for 21 days.
In October 2017, DeepMind’s latest version of AlphaGo paper introduced the strongest and latest version of AlphaGo Zero so far: using pure reinforcement learning, the value network and the strategy network were integrated into a framework, and after three days of training, the previous version of AlphaGo was defeated by 100-0.
7. AlphaGo teaching tools are officially launched. Ke Jie: Re-learn Go.
On December 11th, 2017, Fan Hui, the first teacher of AlphaGo and the champion of European Go, revealed in Weibo: "There are more than 20,000 changes in teaching tools of AlphaGo, consisting of more than 370,000 moves. From the perspective of AlphaGo, this paper analyzes and suggests many moves to start Go. Subsequently, Ke Jie, the first person in Go in the world today, then forwarded this Weibo and commented: "Learn Go again."
AlphaGo has retired, but the technology lives on.
With this article, I would like to pay tribute to AlphaGo and the people who developed AlphaGo.