AI solves Rubik's Cube in 1.2 seconds (that's three times slower than a non-AI algorithm)

www.theregister.co.uk | 7/15/2019 | Staff
joyy (Posted by) Level 3
Click For Photo: https://regmedia.co.uk/2019/07/15/rubiks_cube.jpg

A new neural-network can solve a Rubik’s cube twice as fast as the fastest human – though roughly three times slower than the fastest dumb algorithm – according to research published in Nature Machine Intelligence on Monday.

Though the AI approach is not a nippy as the fastest traditional computational method, specifically the world-beating min2phase algorithm out of MIT, it has promise beyond old-school playground puzzles: for one thing, it could be put to better use studying proteins.

People - Toy - 1970s - World - Cube

People have been fascinated by the popular toy since it was created in the late 1970s. The World Cube Association, an organization focused on combination puzzle games like the Rubik’s Cube, run several competitions every year to challenge fans to solve it as fast as they can. People who enter these contests are known as speedcubers, and the current world record belongs to Yusheng Du, a Chinese speedcuber that cracked the puzzle in just 3.47 seconds.

But Du is probably no match for machine learning software like DeepCubeA. The system developed by a team of researchers at the University of California, Irvine (UCI), made up of a deep neural network can solve a Rubik’s Cube in an average of 28 moves in 1.2 seconds, Forest Agostinelli, first author of the paper and a PhD student at UCI, told The Register.

System - Min2phase - Algorithm - Researchers - MIT

That's fast, but not as fast as the robotic system running the min2phase algorithm developed by researchers at MIT last year. Min2phase, however, doesn't need to be trained, doesn't have a neural network, and doesn't use any machine learning techniques: it's programmed to solve cubes and that's all it can do (and does it extremely quickly).

DeepCubeA was trained using reinforcement learning. Starting with a random combination of the puzzle, it has to find a strategy that will minimize its “cost-to-go-function”, which calculates the cost - or...
(Excerpt) Read more at: www.theregister.co.uk
Wake Up To Breaking News!
Sign In or Register to comment.

Welcome to Long Room!

Where The World Finds Its News!