Search found 81 matches

by Tony P.
Mon Sep 11, 2017 7:43 pm
Forum: Computer Chess Club: General Topics
Topic: Fizbo 1.9 released
Replies: 57
Views: 14928

Re: Fizbo 1.9 released

Thanks so much for this original engine!!! Which program do you regard as the best opponent for Fizbo in terms of the potential for tactical entertainment value? I terminate the game as soon as it starts looking too drawish or balanced :D In most of the below games, I manually put in the first 1-5 m...
by Tony P.
Thu Feb 23, 2017 7:47 pm
Forum: Computer Chess Club: General Topics
Topic: Using GAN to play chess
Replies: 4
Views: 1124

Re: Using GAN to play chess

Curiously, David Pfau and Oriol Vinyals, who're working for DeepMind, have pointed out the close connection between GANs and actor-critic reinforcement learning methods, and the latter have been shown to be efficient at outperforming humans at video games.
by Tony P.
Thu Feb 23, 2017 6:24 pm
Forum: Computer Chess Club: General Topics
Topic: Using GAN to play chess
Replies: 4
Views: 1124

Re: Using GAN to play chess

E.g. if one wants to train the NN to play in the Chigorin style, then one may add a term to the loss function corresponding to an exchange of a knight for a bishop, so that the AI become less likely to give a knight up, as Chigorin preferred knights to bishops. No, my impression was wrong. The GAN ...
by Tony P.
Thu Feb 23, 2017 4:59 pm
Forum: Computer Chess Club: General Topics
Topic: Using GAN to play chess
Replies: 4
Views: 1124

Re: Using GAN to play chess

I've just downloaded the paper and haven't figured out its formulas yet... Evgeniy, could you please give us a brief explanation of what 'style transfer' means? Do I take it right that it means picking up a 'model' human player and setting up a loss function (to use in NN training) in a way that pen...
by Tony P.
Thu Feb 16, 2017 7:36 pm
Forum: Computer Chess Club: General Topics
Topic: Kings Indian expert
Replies: 19
Views: 4244

Re: Kings Indian expert

It's rather obvious that SF is not a fan of the King's Indian (nor the Dutch, for that matter) in general. Black's typical plans in these openings are so longterm that SF doesn't appreciate them until it reaches a big search depth.
by Tony P.
Wed Feb 08, 2017 7:11 pm
Forum: Computer Chess Club: General Topics
Topic: 20 Years Later, Humans Still No Match For Computers
Replies: 18
Views: 4434

Re: 20 Years Later, Humans Still No Match For Computers

corres wrote:In the old times chess play was considered as a kind of Art but now it is only a branch of sport to gain money.
You've made me laugh! Natalia Pogonina estimated in 2010 that only the top 100 players in the world were earning more than $50K a year from chess-related activities.
by Tony P.
Mon Feb 06, 2017 6:36 pm
Forum: Computer Chess Club: General Topics
Topic: Could AlphaGO's concepts be beneficial to chess?
Replies: 7
Views: 2599

Re: Could AlphaGO's concepts be beneficial to chess?

It would be interesting to see the concept that AlphaGO utilized to learn from master GO games applied to chess. I am not a programmer so I am sure it is not that easy or it has been thought of before. To have the concept applied to all wild attacking players or whatever style you want and then hav...
by Tony P.
Mon Feb 06, 2017 3:24 am
Forum: Computer Chess Club: General Topics
Topic: deep sac
Replies: 13
Views: 2668

Re: deep sac

Gull 3 saw a similar line at depth 23/68 (upon processing ~600 meganodes): [pgn][Event "Computer chess game"] [Site ""] [Date "2017.02.06"] [Round "?"] [White "Gull 3 x64"] [Black "Gull 3 x64"] [Result "*"] [BlackElo "2400"] [Time "05:38:21"] [WhiteElo "2400"] [TimeControl "40/120:40/120:40/120"] [S...
by Tony P.
Mon Feb 06, 2017 1:39 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Deep Pink: a chess engine using deep learning
Replies: 20
Views: 6411

Re: Deep Pink: a chess engine using deep learning

it makes sense to design intermediate pseudo-rewards like those for encountering manually programmed (material, open file possession, etc.) or automatically extracted features. This may be an improvement over the simple inclusion of these features into the static eval because log(p) - log(1-p) (whe...
by Tony P.
Mon Feb 06, 2017 1:02 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Deep Pink: a chess engine using deep learning
Replies: 20
Views: 6411

Re: Deep Pink: a chess engine using deep learning

The purpose of the quiescence search, and most of the pruning that happens in the main search, is to cut off rapidly nodes that are losing material. That is pretty necessary because so many parts of the search tree typically involve losing material for no compensation. A hanging piece and a losing ...