Welcome, Guest. Please Login or Register.
Mar 29th, 2024, 12:04am

Home Home Help Help Search Search Members Members Login Login Register Register
Arimaa Forum « Building a MLP evaluation function »


   Arimaa Forum
   Arimaa
   Bot Development
(Moderator: supersamu)
   Building a MLP evaluation function
« Previous topic | Next topic »
Pages: 1  Reply Reply Notify of replies Notify of replies Send Topic Send Topic Print Print
   Author  Topic: Building a MLP evaluation function  (Read 2859 times)
Ikki
Forum Full Member
***




Arimaa player #8506

   


Gender: male
Posts: 16
Building a MLP evaluation function
« on: Jun 21st, 2013, 10:40pm »
Quote Quote Modify Modify

Hello all,
 
I'm trying to build my own evaluation function for an Arimaa bot.
I've got a problem when trying to imagine a good way to make my NN learn.
 
For example, I'd like to be able to provide it thousand of positions with an accurate evaluation in order to make it able to recognize strong position from bad ones.
 
The problem, of course is that i'm not able to provide theses evaluations because it would require me to be an Arimaa GM and even then, it would take decades to provide enough positions to my NN.
 
One of my ideas was to provide my NN positions with something that I would call a "linear evaluation". It would work that way:
 
Imagine a game against A and B. A plays gold and B silver. B win after 50 turns.
 
After turn 0 (setup), the position is even (50/50)
After turn 1 (1 move each side), the position si 1/50 in favor of B (the winner).
After turn 2, it is 2/50 in favor of B.
Etc.
Until turn 50 in which B won and means that B's position worth 100% and A's position worth 0.
 
So it supposes a linear evolution of the advantage of the winning player. It supposes that after 1 turn, the winning player already has an advantage which can be true sometimes but will be wrong most of the time. But maybe that in average, that can be a good method, when provide sufficiently games and positions.
 
Any thoughts about how good or bad this can be ?
IP Logged
Fritzlein
Forum Guru
*****



Arimaa player #706

   
Email

Gender: male
Posts: 5928
Re: Building a MLP evaluation function
« Reply #1 on: Jun 22nd, 2013, 12:14am »
Quote Quote Modify Modify

Have you read up on temporal difference learning?
 
http://en.wikipedia.org/wiki/Temporal_difference_learning
 
It was the idea behind TDGammon, which was, for a time, the best backgammon bot in the world.
 
http://en.wikipedia.org/wiki/TD-Gammon
 
On a cautionary note, make sure to read haizhi's thesis.  His learning bot became superstitious, i.e. the weightings on the learned evaluation were unrelated to good play.
 
http://arimaa.com/arimaa/papers/HaizhiThesis/haizhiThesis.doc
IP Logged

mattj256
Forum Guru
*****



Arimaa player #8519

   


Gender: male
Posts: 138
Re: Building a MLP evaluation function
« Reply #2 on: Jun 25th, 2013, 2:13am »
Quote Quote Modify Modify

A different thread you might find interesting:
 
on Mar 24th, 2008, 9:17pm, IdahoEv wrote:
I have a stored database of all boards that have existed in the game (at least through the last time I ran the updater), meaning every step of every turn of every game saved independently as a board description.
 
I am mining this database to work on the eval function of my bot.   What would be nice to have (but is of course impossible) is some a priori measure of which player is winning  -- and ideally by how much -- that could be used to train an eval function to model that variable.
IP Logged
Pages: 1  Reply Reply Notify of replies Notify of replies Send Topic Send Topic Print Print

« Previous topic | Next topic »

Arimaa Forum » Powered by YaBB 1 Gold - SP 1.3.1!
YaBB © 2000-2003. All Rights Reserved.