The humans won another game and AlphaGo smiled strangely.

Recall that AlphaGo, the originator of the Artificial Intelligence family, survived in the fourth round of the fourth round of the World Champion Li Shishi with humans (now extinct), and survived, leaving behind stars for the subsequent proliferation of artificial intelligence. The fire came from a certain day in 2100.

In the fourth innings of the man-machine war, Li Shishi’s hands were white. After a fierce battle, Li Shishi regained an inning on behalf of mankind. The current score is 3:1. This victory of Li Shishi has saved mankind honor, but some people say: This is a strange smile from AlphaGo.

From the point of view of the experts on the scene, the start of the two sides is similar to the second in the front, and AlphaGo once again has a “Chinese flow” layout. Drew on the experience of a three-game losing streak, Li Shishi's setbacks in the session appeared unusually conservative, and continued to withdraw, the machine quickly gained the upper hand.

Like the first three sets, AlphaGo continues to walk out of the eyes of humans, and Li Shishi dared not be easily disposed of, once again into deep thought. Repeatedly falling into a state of long-term consideration led Li Shishi to take too much time. After the game lasted for two and a half hours, Li Shishi only had 17 minutes, which was one hour less than the remaining time of AlphGo.

For a long time, Li Shishi was in an absolute inferior position in time. The length of time for a long game of chess could reach 15 minutes. However, this unexpected move of the machine did not seem to deceive Li Shishi.

From the results, many thoughts still played a big role. When only about six minutes away from the countdown, Li Shishi took a step in the hinterland, and Li Shishi digged to become a turning point in the game. Followed by step by step attacks, the two again into the stalemate.

AlphaGo constantly "mistakes" and made obvious mistakes in accordance with his "thinking mode." At the time, Hei 97th hand went to the middle of two whites in the lower left corner. Li Shishi made the 98th decision to eat this unjustified sunspot. Experts also It is widely believed that this is Li Shishi's opportunity and the situation is more and more favorable. The next move does not seem to have any suspense, Li Shishi exposed a long absence smile.

At the conference after the end of today's game, Depha Hassabis, the father of AlphaGo and head of Google DeepMind, expressed this kind of "fault" to 78 hands. This may be due to AlphaGo's algorithmic flaw. Computers exist in extremely complicated fighting situations. Misunderstandings and blind spots.

Through this game, Li Shishi seems to have discovered the weakness of AlphaGo. One of the great possibilities is that AlphaGo's decision-making network may not have calculated the 78th-hand data of Li Shishi, making the situation more complicated, which has led to later. The judgment crash caused a "Bug".

According to the analysis of the first three games by Mr. Chen Jing, master of computer science at the Hong Kong University of Science and Technology (the first three sets of replay analysis focused on the "recovery" of Titanium Geek's WeChat reply), AlphaGo has advantages and potential defects relative to humans. Coexistence:

First of all, its overall situation is much stronger than humans, because there are powerful computing resources to ensure that the final number of simulations is adequate, and pruning of strategic networks and value networks ensures the quality of simulations. It will use a small window in the closed part of the killing to be exhaustive and will not lose.

It has a good layout and strong control of the battle in the midrange. It is a good example of the overall situation. It is almost closed in the middle and late rounds. It is basically exhaustive, the calculations are very accurate, and everything is simulated after millions of simulations. It's not a question of wanting to end up in the game, it's a win, but it's impossible to end the reversal, it affects the winning percentage and it immediately blocks you back.

However, AlphaGo may have the possibility of inaccurate depreciation in the small means of a closed situation and do not know how to reason. In open contact warfare, if the battle is going to go very far, it may also be too numerous to count and reveal flaws. But it will not be a collapsing flaw. If it is to collapse, it will surely know that it has been lost. If it does not collapse, it may be confused. At present, it is such two minor problems.

As a result, Li Shishi has the possibility of winning. However, AlphaGo suddenly lost control in the game today. Dr. Huang Shijie, who replaced AlphaGo's son, was also very puzzled at the time. Then there is no water in the game today. Is it possible?

All kinds of speculations sprang up in a flash, and some comments jokingly said, “Does the dog win the podium and the process begins to pride?”, “This few steps is obviously inferior, you are tired!” and even Some netizens suspect that Google is not releasing water.

And it seems quite reasonable to worry about it. "AlphaGo's algorithm is not open to the public. What kind of step is the merits? The outside world has no way of knowing that this game is Google's typical publicity game to appease Korea and the world. Few people are concerned about the papers of the DeepMind team for several years. In fact, their algorithm is much more powerful than the outside world has imagined.”

Hearing this point of view, it is really a matter for the titanium geeks to make up their minds and think about it! If AlphaGo really loses it intentionally, then the ending is a bit daunting!

Google is a great company. This game, or that this research has brought humanity to a new height, from this point of view, whether the result of the human-computer war is 4:1 or 3:2 doesn't seem to have much to do with it. AlphaGo, as the representative of Google artificial intelligence, has defeated the World Go Championship twice in a row and has greatly improved its learning ability within five months.

The fact that humanity has been defeated in the game of Go has been quietly underway. Even if Li Shishi continues to win the next game, it will only continue to be a sample of AlphaGo to make it stronger. In turn, it will sooner or later surpass humanity. thing.

Watching everyone in this century’s showdown seems to underestimate the pace of development of artificial intelligence (AI). The truth is that machines have really made us amazing. Behind the victory today, it’s not a bad idea. Fear, the conspiracy to control the machine, the artificial intelligence that is awakening to self, to all unknown things.

Photovoltaic Installation Tools

Pv Testers,Photovoltaic Installation Tools,Photovoltaic Tester,Solar Pv Tester

Sowell Electric CO., LTD. , https://www.sowellsolar.com