Author |
Topic: Announcing Development of bot_tod (Read 955 times) |
|
maker
Forum Full Member
bot_tod's maker
Gender: 
Posts: 21
|
 |
Announcing Development of bot_tod
« on: Sep 23rd, 2004, 6:07pm » |
 Quote  Modify
|
Hey guys,
I just thought I'd let you guys know the name of my soon(2->3 months) to be released bot: "drumroll please..." bot_tod!!!  I'm not much of one for cool names, so I thought I'd do the obvious. Anyway, the idea behind my bot is that the poor  thing will try to 'learn'  how to play the game from the masters: you. I hope that it will do ok. Anyway, back to implementation of said bot... 
maker
|
|
 IP Logged |
|
|
|
99of9
Forum Guru
Gnobby's creator (player #314)
 
Gender:
Posts: 1413
|
|
Re: Announcing Development of bot_tod
« Reply #1 on: Sep 23rd, 2004, 7:14pm » |
Quote Modify
|
Now that the historical game database is available, this is finally truly possible.
My bot already tries to learn the weights in its evaluation function from past games it has played, but I haven't had a lot of success yet (ie my hand weighted bot is better). I think learning from the entire history of arimaa might help! But I'm yet to try implementing that.
Good luck with your bot.
|
|
IP Logged |
|
|
|
maker
Forum Full Member
bot_tod's maker
Gender:
Posts: 21
|
|
Re: Announcing Development of bot_tod
« Reply #2 on: Sep 23rd, 2004, 10:45pm » |
Quote Modify
|
Thanks for the encouragement. You are right, I wouldn't be able to do this without a database to use as a 'teaching tool' for bot_tod; however, I plan on using it selectively, so that the bot will learn from whom and what games I believe will teach it the most. Not that I'm that great at learning from my past mistakes(see my record against bot_Arimaazilla), but once I've written the final learning algorithm, I believe I'll know how to make it the strongest it can possibly be. Of course there's always tweaking...  Who could imagine such a thing?
maker
|
|
IP Logged |
|
|
|
|
Home
Help
Search
Members
Login
Register
Reply
Notify of replies
Send Topic
Print