Diadem comments on [LINK] Deep Learning Machine Teaches Itself Chess in 72 Hours - Less Wrong

8 Post author: ESRogs 14 September 2015 07:38PM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (15)

You are viewing a single comment's thread.

Comment author: Diadem 15 September 2015 10:50:07AM 6 points [-]

Straight out of the box, the new machine plays at the same level as the best conventional chess engines, many of which have been fine-tuned over many years. On a human level, it is equivalent to FIDE International Master status, placing it within the top 2.2 percent of tournament chess players. But even with this disadvantage, it is competitive. “Giraffe is able to play at the level of an FIDE International Master on a modern mainstream PC,” says Lai. By comparison, the top engines play at super-Grandmaster level.

That's a pretty hard contradiction right there. The latter quote is probably the correct one. Modern chess engines beat any human player these days, even running on relatively modest hardware. That's assuming full length games. At blitz games computers are much better still, compared to humans, because humans are much more error prone.

So if this neural net is playing at master level it's still much, much weaker than the best computers. From master to grandmaster is a big leap, from grandmaster to world top is another big leap, and the best computers are above even that.

Still interesting of course.

Comment author: TheMajor 15 September 2015 07:12:59PM *  2 points [-]

Yes, I noticed this too. The paper itself compares Giraffe (ELO ~2400) to 8 other chess engines (page 25 of the PDF), and decides that

It is clear that Giraffe's evaluation function now has at least comparable positional understanding compared to evaluation functions of top engines in the world.

For comparison, a frequently updated list of chess engines and their (approximate) ELO ratings, which would list Giraffe around shared 165'th place. It seems that it is the reporting, instead of the paper, that is making the exaggeration.