machine one, zero man. It ended well the first historic match between AlphaGo software and world champion hardest game ever invented, the Go. Follow other four clashes from here until March 15, also in Seoul. The ending is not yet written but everything leads one to believe that it is mortgaged. Hard not to sympathize with the Korean Lee Sedol, whose talent human yesterday had to surrender digital intuition, and now will try to win the second round. Impossible not to cheer for the ingenuity of the “computer science specialists” that the DeepMind (Google subsidiary) have created a system capable of beating the best players. The stakes are highest prize of one million dollars that would go to charity if the famished AlphaGo. The victory of the software would have a strong symbolic value and may represent a breakthrough in artificial intelligence.
odds
it had already happened that the sconfiggessero computer checkers samples, backgammon and chess. But the ancient game of Go was considered the most difficult to conquer the fortress, at least until January Nature has not announced that the DeepMind had beaten the European champion and was ready to repeat the exploit with the most good of the globe. The Go checkerboard is a grid with 19 horizontal and 19 vertical positions (instead of eight for eight as in chess) and the number of configurations is equal to a power of ten with 170 as the exponent (the consideration for chess is a modest 10 to fiftieth). The number of possibilities is amazing, therefore, greater than the sum of the atoms in the universe, too great an algorithm can find the best move in a comprehensive way. No wonder that with the pieces of this pastime, popular among China, Japan and South Korea, have played the most brilliant minds, from Albert Einstein to Alan Turing, John Nash to Paul Erdos. The brute force of the calculations is not enough, the Go requires flexibility, timing, a balance between attack and defense. quality and most mysterious of chess, his philosophy is to win a territory rather than eliminate the opponent’s pieces, and this made it attractive also for literature, with quotations ranging from The Elegance of the Hedgehog Muriel Barbery to Satori Don Winslow.
the medical applications
When the chess player Garry Kasparov was defeated by Deep Blue was a shock, but in the meantime the AI is not It took off as hoped. Why then this would be a turning point? The IBM computer had been programmed to win a specific game, instead AlphaGo is more generalist and learned from trial and error, considering a 30 million database of moves made by players in the flesh and then challenging himself. Do not waste time evaluating unlikely options, it focuses on the most promising. To achieve this uses two neural networks inspired by the brain, where connections are strengthened by the experience. These features make it more versatile and – hopefully – suited to solve real problems that require recognition of complex patterns and long-term planning. In the future could use medical images to diagnose and recommend treatment plans, or improve the models underlying the forecasts on climate change, suggests DeepMind. Compared to the challenge with the European Fan Hui champion, who had ended 5 to zero for silicon intelligence, the system seems improved. Sedol admitted he made a mistake at the beginning of the game that has conditioned the entire match and forced him to give up after three hours and a half. For the rematch today, however, he is confident. This time to open will be the car with the black pieces, will touch the white man.
9 March 2016 (edited March 9, 2016 | 23:29)
© ALL RIGHTS RESERVED
No comments:
Post a Comment