wfolsom/msproj
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

First working implementation of ANN which is trained using GameResults.

Browse Source 
The PassFilter simply outputs BlackWins and WhiteWins (Range 0 - 1 but not presently clamped).
In principle, this type of feedforward ANN can be used to decide whether a PASS will result in blackwins or whitewins at any stage.
The goal is for the network to learn that passing while losing when valid moves exist is bad, but passing while winning is relatively harmless later in the game.
This commit is contained in:
Woody Folsom
2012-11-17 18:40:31 -05:00
parent d9d6ecda80
commit aca8320600
37 changed files with 1040 additions and 544 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
data/tourney1/gogame-121117181002-0500.sgf Normal file
View File

@@ -0,0 +1 @@
(;FF[4]GM[1]SZ[6]KM[1.5]RE[W+26.5];B[ab];W[ff];B[ca];W[df];B[de];W[ce];B[cf];W[ba];B[aa];W[eb];B[be];W[bf];B[dc];W[ea];B[bc];W[cc];B[dd];W[fc];B[ad];W[db];B[fd];W[fa];B[cb];W[ed];B[];W[ec];B[fe];W[af];B[cd];W[ef];B[];W[ae];B[];W[bd];B[];W[ac];B[bb];W[da];B[ad];W[ee];B[];W[cc];B[dc];W[dd];B[];W[ac];B[];W[fd];B[];W[ba];B[aa];W[cb];B[];W[ab];B[];W[bc];B[];W[])