Skip to content

Releases: vaquierm/Pentago-Swap-AI-Agent

Monte Carlo Tree Search 3.0

19 Mar 03:41
15619f6
Compare
Choose a tag to compare

Added evaluation function to reward blocks against the opponent
Added logic to always play in around the opponent when they start to start off with a very defensive game

line 569 in BoardStateClass should be antiPatternPresentOpponent

Monte Carlo Tree Search 2.0

17 Mar 16:03
Compare
Choose a tag to compare

Montecarlo with a slight bias for choosing move based on board evaluation function

Monte Carlo Tree Search 1.0

16 Mar 22:02
Compare
Choose a tag to compare

Does pretty good.
Can be beaten if it is cornered but in general does very well

Monte Carlo Tree Search v0.3

16 Mar 16:17
Compare
Choose a tag to compare
Pre-release

Dumb things have been fixed. Its now actually beating randomplayer.
Could consider event less moves to see the performance

Monte Carlo Tree Search v0.2

16 Mar 03:11
e2f187c
Compare
Choose a tag to compare
Pre-release

Can beat random player.
Humans can still easily beat it.
Should figure out a block mechanic

Monte Carlo Tree Search v0.1

15 Mar 16:31
Compare
Choose a tag to compare
Pre-release

Monte Carlo tree search implemented with upper confidence tree.
The main issue is that the branching factor is so large that the first few moves seem pretty random