Search found 782 matches

by matthewlai
Wed Oct 23, 2019 5:21 pm
Forum: Computer Chess Club: General Topics
Topic: Poll: How Many "Weights" Needed To Play "Known" Chess Very Well?
Replies: 34
Views: 1810

Re: Poll: How Many "Weights" Needed To Play "Known" Chess Very Well?

A few things - 1. Saying an NN is equivalent to a linear system of equations is not quite true. It's true for the very first "neural networks" from the 1960s and 1970s, but that's actually what stopped them from progressing - they cannot model any non-linearity, and most real world decision functio...
by matthewlai
Wed Oct 23, 2019 4:39 pm
Forum: Computer Chess Club: General Topics
Topic: Poll: How Many "Weights" Needed To Play "Known" Chess Very Well?
Replies: 34
Views: 1810

Re: Poll: How Many "Weights" Needed To Play "Known" Chess Very Well?

A few things - 1. Saying an NN is equivalent to a linear system of equations is not quite true. It's true for the very first "neural networks" from the 1960s and 1970s, but that's actually what stopped them from progressing - they cannot model any non-linearity, and most real world decision function...
by matthewlai
Wed Aug 28, 2019 12:09 pm
Forum: Computer Chess Club: General Topics
Topic: OpenSpiel: DeepMind's new generic games library
Replies: 5
Views: 1513

Re: OpenSpiel: DeepMind's new generic games library

No checkers or other draughts-like games? :x Contributions warmly welcomed :). We had limited time and wanted to implement a wide range of games from different categories, so some popular games have not been implemented. Would it be straightforward to implement an AI that plays to tetris? Yes! Tech...
by matthewlai
Tue Aug 27, 2019 5:28 pm
Forum: Computer Chess Club: General Topics
Topic: OpenSpiel: DeepMind's new generic games library
Replies: 5
Views: 1513

OpenSpiel: DeepMind's new generic games library

GitHub: github.com/deepmind/open_spiel Paper: https://arxiv.org/abs/1908.09453 The library has about 25 games implemented (including chess and Go), all using the same interface, so if you write an AI that plays games using that interface, it will be able to play any of those games (of course, writin...
by matthewlai
Thu Mar 28, 2019 11:32 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Some x64 assembler for the curious
Replies: 21
Views: 5521

Re: Some x64 assembler for the curious

I have already done the proof that handwritten assembler is faster. I have programmed two perft examples. One in pure C and one with handwritten assembler for the move generator, make move and take back. Both versions make and unmake all moves generated and do no cache counting. GNUChess 4 style in...
by matthewlai
Tue Mar 26, 2019 3:22 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Some x64 assembler for the curious
Replies: 21
Views: 5521

Re: Some x64 assembler for the curious

; ATTACKED BY BLACK - rax contains the square in question Atkbyblk proc mov r8d, wp[rax*4] ; wp indexed by square gives first entry in huge move table abb1: mov r9d, tosq[r8*4] ; from tosq[] we get the first destination square mov r10d, [rcx].t.board[r9*4] ; the index of piece on board - empty squa...
by matthewlai
Sat Jan 05, 2019 7:30 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: mcts question
Replies: 2
Views: 741

Re: mcts question

A: After tree policy, I am at a leaf node (for my tree in memory) and I want to expand. I figured a good idea would be expanding all children in parallel, but when you expand more than one child visit counts become an inaccurate way to judge the exploration of a given node because leaves have diffe...
by matthewlai
Tue Jan 01, 2019 2:47 am
Forum: Computer Chess Club: General Topics
Topic: Recent Alpha zero vs Stockfish 8 match.
Replies: 41
Views: 5635

Re: Recent Alpha zero vs Stockfish 8 match.

First some terminology: a value function is a function that predicts the outcome of the game given an input state. A policy function is a function that predicts the best move given a state. They are both necessary in both alpha-beta and MCTS, but that's more a practical detail than theoretical. The...
by matthewlai
Mon Dec 31, 2018 6:33 am
Forum: Computer Chess Club: General Topics
Topic: Recent Alpha zero vs Stockfish 8 match.
Replies: 41
Views: 5635

Re: Recent Alpha zero vs Stockfish 8 match.

NN keeps the results of the self play. When you use an NN-type engine the engine does nothing but search for sequence of moves in NN. Do you think Stockfish also uses pre-evaluated positions? Stockfish and other AB-type engines use a type of evaluation and NN-type engines use an other type of evalu...
by matthewlai
Mon Dec 31, 2018 6:29 am
Forum: Computer Chess Club: General Topics
Topic: Recent Alpha zero vs Stockfish 8 match.
Replies: 41
Views: 5635

Re: Recent Alpha zero vs Stockfish 8 match.

First some terminology: a value function is a function that predicts the outcome of the game given an input state. A policy function is a function that predicts the best move given a state. They are both necessary in both alpha-beta and MCTS, but that's more a practical detail than theoretical. Theo...