AlphaGo seals 4-1 victory over Go grandmaster Lee Sedol
DeepMind’s artificial intelligence astonishes fans to defeat
human opponent and offers evidence computer software has mastered a major
challenge
Google DeepMind’s AlphaGo program triumphed
in its final game against South Korean Go grandmaster Lee Sedol to win the
series 4-1, providing further evidence of the landmark achievement for an
artificial intelligence program.
Lee
started Tuesday’s game strongly, taking advantage of an early mistake by
AlphaGo. But in the end, Lee was unable to hold off a comeback by his opponent,
which won a narrow victory.
After
the results were in, Google DeepMind co-founder Demis Hassabis called today’s
contest “One of the most incredible games ever,” saying AlphaGo mounted a
“mind-blowing” comeback after an early mistake. This was the fifth
game in seven days, in what was a draining, emotional battle for Lee. AlphaGo
had won the first three, but Lee took the fourth game on Sunday.
He
remained in his seat as the game’s results were announced, his eyes swelling
with tears. In a post-game press conference, he expressed regret over his
defeat. “I failed,” he said. “I feel sorry that the match is over and it ended
like this. I wanted it to end well.”
Throughout
the match, Lee won praise from observers for a determined, creative approach to
AlphaGo, an opponent that is invulnerable to stress and fatigue. In Tuesday’s
press conference, Chris Garlock, one of the live commentators said the match
was composed of “five beautiful and historic games,” adding, “I think we’ll be
studying these for years to come.”
Due to
Go’s complexity and the importance of reaction and intuition, it has proved
harder for computers to master than simpler games such as checkers or chess. Go
has too many moves for a machine to win by brute-force calculations, which is
how IBM’s Deep Blue famously beat former world chess champion Garry Kasparov in
1997.
AlphaGo’s
win over Lee is significant because it marks the first time an artificial
intelligence program has beaten a top-ranked Go professional, a victory experts
had predicted was still years away. AlphaGo beat European Go champion Fan Hui
in October, but Lee was expected to be a tougher challenge.
The match
has brought an unusual level of attention to Go, a game that is popular in east
Asia but not widely played in the west. Go insiders say they are not used to
being in the spotlight. “I’ve never seen this much attention for Go, ever,” Lee
Ha-jin, secretary general at the International Go Federation and guest
commentator on Tuesday’s live broadcast, said.
Google
DeepMind has talked about applying the deep neural networks and machine
learning techniques that AlphaGo used to master Go to more pressing areas such
as healthcare and robotics. But with AlphaGo’s victory in the books, Hassabis
was tightlipped, saying his team will need to return to the UK and spend
“weeks or months” going over the results of the match before announcing their
next moves.
Stucture of the Lead:
Who:AlphaGO and GO grandmaster Lee Sedol
When:In March 2016.
What:AlphaGO defeats GO grandmaster Lee Sedol.
Why:Not mentioned.
What:AlphaGO defeats GO grandmaster Lee Sedol.
Why:Not mentioned.
Where:Not mentioned.
How:AlphaGO mounted a ststem which could let it learn the mistakes and correct them. Therefore it can defeat GO grandmaster.
Keywords:
1.mounted 安裝
2.draining 枯竭
3.swelling 腫脹
4.commentators 評論員
5.intuition 直覺
6.brute-force 強力的
7.calculations 計算
8.robotics 機器人
9.tightlipped 守口如瓶的