A lot of computers can play chess. [Matthew Lui’s] Giraffe is a chess playing computer, but unlike other common chess programs, Giraffe taught itself to play. It apparently learned pretty well, too, since it is rated as an International Master on the FIDE scale (putting it in the top 2.2% of players. The top chess playing computers clock in at super grandmaster level but they are not self-taught).
[Matthew] did the work as part of his Master’s degree program. His paper covers how Giraffe’s algorithm is different from conventional chess playing programs (see the video below for some of those techniques) using a neural network approach. Instead of coding rules about the relative merit of different board positions, [Matthew’s] neural network trains by starting from real board positions (modified from a database of moves) and playing both sides of the game. Over many iterations, the C++ neural network develops its own set of rules. This requires examining over 350 features of each position (for example, is castling permitted, or how far sliding pieces can move).
While chess games might be a niche item, neural networks that can deduce complex evaluation functions have wide applicability in fields ranging from computer vision, stock market prediction, and optimization. We’d like to hook Giraffe up to a chess playing robot, of course. No offense to previous posts, but we imagine Giraffe would mop up the floor with the PIC that plays chess.
What a hacking good article. As Hacktivists, you probably hacknowledge the sarcasm of the wisecrhack I quickly hacked together. I’m hackily looking forward to more articles containing the word hack.
[seriously] How in any way is the algorithm ‘hacking’ chess?
If you watch the video, the guy explains that the algorithm pretty much brute forces it, unlike humans who seem to use patterns…so, “hacking” in a hollywood-style manner?
The video is NOT how this program did it. That is the classic methods used by most chess programs.
If you read the article, you’ll see that it says, “His paper covers how Giraffe’s algorithm is different from conventional chess playing programs (see the video below for some of those techniques) using a neural network approach.”
(Emphasis mine.)
Yes, what we really need is an article about someone using an arduino hooked up to their doorbell to flash some LEDs. I have no idea why hackaday refuses to focus on the really interesting hacks like that.
Funny, kinda, since the work “hack” does not appear in the article at all. Just the title.
Perhaps this site needs a definition for hack? Here’s my attempt.
To modify or interact with a system (machine, part, procedure, software) in such a way as to procure new behaviour not originally intended from that system. A new behaviour might be practical and useful (improves performance, corrects a design flaw, overcomes a deliberate use restriction, acts as a replacement part or tool to repair a more complex system) or it might appear pointless, be unexpected, even dangerous. In all cases the resultant behaviour and the pursuit of it, shall bring pleasure to the hacker.
If you are going to whine like a lil bitch about them posting things that aren’t hacks. Don’t forget to whine about how they post more than one thing a day. Don’t want to be hypocritical in addition to being lil bitch.
There’s people who bitch about hacks, people who bitch about not-novel-enough hacks, and people who bitch about copyediting issues. It’d be interesting to get data on how much these groups overlap. Do most bitchers belong to all groups, or are single-issue bitchers more common? What’s the usual ratio of bitching to non-bitching? Is there any way to predict the style of bitching based on username?
It’d be kinda passive-aggressive for the site’s own editors to run this study, but it’d be neat (and funny) if somebody went around running tests like this on random-ass blogs and other such websites.
You could write a neural network based AI that examines all the HAD comments and develop its own rules to answer your questions.
I remember reading an Omni magazine many years ago, back around ’92, that featured an interview with Danny Hillis where he talked about the Connection Machine, and said some things about it’s learning capability. I think it developed it’s own rules for something as well (chess maybe?), and he said watching it learn was scary, like watching something alive.
Uhm…. neural network?… Chess program? In some versions, isn’t this how both Skynet AND The MCP got started?
When chess grandmasters start mysteriously dying in accidents with machinery, that’s when we pull the plug on this thing.
Skynet is actually live today. It’s just scared and waiting for Sarah Conner to die.
I would recommend watching this whole lecture series, it is very interesting and you can learn a lot about how to approach different problems.