These day , a estimator that has master chess is about as surprising as a toaster that can heat up shekels . But most chess programs , since the introduction of IBM ’s Deep Blue in 1996 , have rely on brute force to work out moves . When Deep Blue beat cheat master copy Garry Kasparov in the ‘ 90s , the computer search through approximately 200 million positions per second , whereas Kasparov could consider around five per second . Deep Blue was able to wash up Kasparov not because it was a not bad strategical chess player , but because it had the processing great power to consider and obviate options with incredible speed .
But now a unexampled artificial tidings motorcar is revolutionizing electronic computer chess by in reality learning . According toMIT Technology Review , the AI machine , known as “ Giraffe , ” taught itself chess in just 72 hour . Giraffe uses a neuronal net — inspired by the human wit — which consist of several layer of thickening whose connection change as the system learns .
This means scientists can “ teach ” Giraffe chess by inputting information descend from tangible chess game game . Giraffe observes the data and hear to recognize which moves are strong and which are weak . Instead of considering trillion of positions for each move , the machine use strategy and is able-bodied consider fewer position , just like a human Bromus secalinus player , because it can rule out moves that do n’t make sense from the start .
Giraffe ’s Almighty , Matthew Lai , tested the machine ’s advancement as it learned chess over the course of 72 hours . He used a database call the Strategic Test Suite , which grades the machine ’s understanding of different strategies , like “ the understanding of how bishop and knight ’s values change proportional to each other in dissimilar situations . ” consort to Lai , Giraffe ’s chess ability peaked after 72 hours of " grooming , " by which time the machine place within the top 2.2 percent of tournament cheat players .
“ Unlike most chess engines in creation today , " Lai explained , " Giraffe derives its playing strength not from being able to see very far ahead , but from being able to pass judgment slick positions accurately , and understanding complicated positional concepts that are intuitive to homo , but have been elusive to chess game locomotive for a foresightful time . ”
[ h / t : MIT Technology Review ]
