قالب وردپرس درنا توس
Home https://server7.kproxy.com/servlet/redirect.srv/sruj/smyrwpoii/p2/ Technology https://server7.kproxy.com/servlet/redirect.srv/sruj/smyrwpoii/p2/ AI learns to solve the Rubik's cube for 1.2 seconds

AI learns to solve the Rubik's cube for 1.2 seconds



However, UC Irvine's DeepCubeA does not have a record for automated Rubic Cube solving. Last year, researchers built a robot that could complete the puzzle for 0.38 seconds. The Min2phase algorithm of the Massachusetts Institute of Technology, which is not a system of artificial intelligence, solves it three times faster than DeepCubeA. While other methods were specifically designed to solve the cube, DeepCubeA had to build their own path. each of her six faces. There are billions of possible combinations for the cube, but only one completed state. While scientists have shown to the AI ​​what the final result looks like, DeepCubeA had to figure out how to get there and they still have no full understanding of how they have developed their strategies.

Researchers started with a simulated version of Rubik's graduate cube, then mixed it up. DeepCubeA was then trained to solve the puzzle for two days, improving its skills while trying to make increasingly difficult combinations. According to an article published in Nature researchers have provided 1

0 billion combinations of DeepCubeA and forced it to solve the puzzles in 30 or fewer strokes. He cracked the puzzle each time and did it with the minimum number of moves in about 60% of the attempts. The algorithm can also find solutions for other games, including Slide Puzzle, Lights Out and Sokoban. by discovering patterns and theorizing with a small human contribution. He adopts a reinforcement approach through which he has learned "how to tackle increasingly difficult states in the opposite direction of the state's goal without any specific knowledge of the field." Repeat. While this also resolves Rubik's cube every time, it usually does so in 30 moves with an average resolution time of 10 minutes. have wider implications. "How can we create a sophisticated AI that is smarter, healthier, and capable of thinking, understanding, and planning?" Professor of Computer Science and senior author of the study, Pierre Baldie said in a statement. "This work is a step towards this grave goal."


Source link