To submit your final each team should:

• Post a final report PDF document (as a comment to this page).
• Submit by email a zip file containing your bot source code and binary (following the same directions as for Project 1, but following them more closely than most teams did for Project 1!). You can also post your bot or a link to your bot on the site (but it is not required to do so).

Before Saturday, 14 May at 4:59pm each individual should send me an email with answers to the following questions:

1. Are you interested in continuing to work on this towards a bot to submit to the ACPC on July 1?
2. If so, explain what time commitment you would want to make to this (i.e., hours per week from not until July 1).
3. How interested are you in attending the ACPC at the AAAI 2011 conference (August 9-10 in San Francisco)?
4. If you want to attend it, do you have your own source of funding or would you need funds from me?
5. Would you be interested in presenting a poster at the poster session for the ACPC? (Note: we need to notify them of interest in this by May 15.)
6. Are you interested in participating in “home games” during the summer?

In this online article, you will have a lot source code for poker bots. You may use them as benchmark players to play against.

http://www.ke.tu-darmstadt.de/resources/poker

These bots are written in Java and require the Pokerserver Frame work

Anyway, we can learn from their source code.

I decide to talk about adaptive exploitive poker bots on Thursday.

I will be focused on AKI Real Bot (and VexBot, optional), 2 very good exploitive bots.

Here is the bibliography:

An Exploitative Monte-Carlo Poker Agent.

http://www.ke.tu-darmstadt.de/publications/reports/tud-ke-2009-02.pdf

These 2 papers talk about an exploitive bot that won the 2nd place in AAAI-08 Computer Poker Competition (i.e. ACPC 2008).

http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.133.5068

This paper talks about VexPlayer, another exploitive bot.

I am also looking for binaries of the winning bots in the ACPC. Hope I can find some so that we can use them as benchmark players to play with.

Hui Shu will lead Thursday’s class, and will post the topic for that class soon.

On Tuesday we will be discussing ways to compute epsilon-equilibria for poker. We’ll review the CFR method previously discussed by Samee, then expand on the technique used in the Gilpen paper presented Thursday by Weikeng. The main focus of this class will be on convex optimization and how it can be used with Nerterov’s excessive gap technique to find epsilon-equilibria in poker. Papers to be covered include:

EDIT: Papers updated!

http://www.cs.cmu.edu/~gilpin/papers/egt.optimizationonline08.pdf

http://www.cs.cmu.edu/~gilpin/papers/iterated.aaai08.pdf

http://www.cs.cmu.edu/~gilpin/papers/egt.wine07.pdf

www.cs.cmu.edu/~gilpin/slides/nesterov.theorylunch07.ppt

-------- Original Message --------
Subject: 	Re: Dealer bug
Date: 	Mon, 4 Apr 2011 15:56:36 -0600
From: Nolan Bard 
To: 	David Evans 

Hi David,

... effectively there was no reason for the return values
of isValidAction to be {-1,0,1}.  A return value of 0
indicates that the action is invalid, non-zero now
indicates a valid action.
...