Difficulty is called the AI ​​challenge of the Apollo program, and it was finally won by a machine "dog." Its deadly magic is to imitate the "deep learning" working principle of human beings. How does the machine learn like a human being? One day it is omnipotent, what about the future of mankind? The author interviewed domestic experts for interpretation.
The principle of Alpha Dog is "deep learning"
AlphaGo was developed by DeepMind, a Google-based company based in London, England. Its main working principle is "deep learning."
In the face of any chess, an intuitive and lazy idea is to violently enumerate all the winning schemes that will form a tree map. However, it is not easy to make it possible for a computer to use this method to defeat humans in the prescribed time, because there are too many possibilities for Go.
There is a saying that the number of possibilities for Go is much greater than the total number of atoms in the observable universe. At the same time, the law of Go is very subtle. To some extent, the choice of the game depends on the "intuition" formed by the accumulation of experience. Therefore, the Go challenge is called the Apollo program of artificial intelligence. This time, the researchers finally came up with the ultimate weapon - "deep learning", allowing the machine to imitate the way that high-end players play chess. Machines can defeat humans by learning humans.
Liu Zhiqing, a professor at Beijing University of Posts and Telecommunications and a director of the China Artificial Intelligence Society, said in an interview with the Beijing Morning Post that Alpha Dogs learned all the high-level Go chess games through the "neural network", which is probably about 200,000 professional chess records in history. Gained the intuition of how to drop the board on the disc. "Assume that we have seen a lot of cats. When we look at the pictures, we judge whether there are cats on the pictures through the first intuition. This kind of intuition is not born, but learned through learning." Liu Zhiqing said.
"So, is the killer of the Alpha dog really the intuition and the sixth sense?" In response to such doubts, Liu Zhiqing said that the intuition it acquired is different from that of human beings. There is a solid mathematical model behind it, and the computer has done a lot. The calculation, in a large number of suspense to find a winning rate, which is not exactly the same as the human brain's intuition, is a manifestation of mathematics, science, human beings can not be achieved.
Aiming at the medical field to recognize cancer cells
Liu Zhiqing said that similar deep learning has appeared in recent years, and some domestic and foreign companies and research institutes are also doing it, but Google’s current work is undoubtedly the best. At present, this technology has some applications. The simplest example is to identify cats through deep learning. Through this identification and verification, more practical applications have been introduced, such as identifying whether there is cancer cells in a certain picture, whether there is wear and tear on the track along a certain railway, or even in military operations, whether the other party’s line of sight is in sight. There are tanks that can be realized through deep learning. Google's automatic driving, which is very important is to identify roads, traffic lights, road signs, etc., which are obtained through deep learning.
Google also expressed its belief that one day, in the future, artificial intelligence technology will be used to solve many real-life problems, such as the establishment of complex climate change models and infectious disease analysis.
In the fourth game of the Alpha Dog and Go Champion Li Shishi in the fourth game yesterday, Li Shishi’s 78th hand was praised as “one of the godsâ€. After this move, the Alpha dog's program was suspected to have appeared "BUG", and it has successively issued several tricks "stinky chess" and lost the game. After the game, everyone cares that the Alpha dog will develop into the medical field in the future. Will this "BUG" also appear? DeepMind founder Hassabis said that the Alpha dog is a preliminary program, not an upgraded version, so it is necessary to look at the flaws, and hope that Li Shishi can help find the shortcomings to further improve. At present, Alpha Dog is still only a procedure, has not yet entered the medical field, and will conduct more rigorous testing.
With deep learning, what will happen to the intelligent age of the future? Liu Zhiqing said that there are many things in the future, people can be replaced, driving planes, cars, operating machine tools, translation and so on. But creativity is a unique feature of human intelligence. In this regard, whether future machines can transcend humanity is still unknown in academia.
Clinic Nitrile Gloves,Nitrile Gloves Small,Medical Gloves Unsterilized,Nitrile Gloves Powder Free Latex Free
Puyang Linshi Medical Supplies Co., Ltd. , https://www.linshimedicals.com