On Tuesday April 5th Colin Braley and Samee Zahur will present in class. The title of the presentation is:
“Mathematical Tools for Machine Learning and Opponent Modeling”
We hope for this class will be more of a generic lecture/discussion on interesting algorithms and mathematical tools in machine learning. While specific applications to poker opponent modeling will be emphasized, we hope to cover these topics in a general manner as well. This structure has been chosen for three reasons. First, many previous classes have discussed a single paper and a break from the norm may be refreshing. Secondly, good poker research papers are in short supply. Third, we hope that a general discussion could lead people to form totally new ideas for their poker bots, which would be great.
This lecture will be largely self contained. However, it will be very helpful to be familar with the following topics:
- Eigenvalues and Eigenvectors (can be found in any linear algebra text)
- Support Vector Machines (covered in class previously)
- Creating an SVM to Play Strong Poker
- Basic probability and statistics (in particular Covariance)
Tentatively, we hope to cover some subset of the following topics:
- Principal Component Analysis(PCA) for use with SVM’s and for opponent modeling
- K-Nearest Neighbors for Opponent Modeling
Code for PCA now available at:
https://subversion.assembla.com/svn/pca_uva/
You can check it out into the current directory via:
svn checkout https://subversion.assembla.com/svn/pca_uva/trunk .
Recent Comments