Artificial Intelligence Reading Group

Welcome to the Artificial Intelligence Reading Group homepage. We are a group of graduate students that meet approximately every other week to discuss papers relevant to our research in Artificial Intelligence and Bioinformatics.

Papers discussed can range from the most recent conference proceedings to AI classics to key papers on the AI qualifying exam list. The goal of the reading group is to give members a a chance to see what other researchers in the department's AI area are thinking about and working on.

We will also host preliminary research talks, alternating with our usual paper presentations, where students can present and discuss their current work and ideas outside of a formal paper presentation. This alternating week schedule may also be used for conference practice talks.

Anyone interested in participating in the reading group should subscribe to the airg mailing list. Instructions for subscribing:

https://lists.cs.wisc.edu/mailman/listinfo/airg

Meetings in Spring 2009

Meetings will be at 4:00pm on Wednesdays in the Computer Sciences building (room 3310). This is the current schedule of presentations:

Subscribe: RSS | iCal

February 4, 2009: Paper presentation by Gautam Kunapuli:
K.P. Bennett, Jing Hu, Xiaoyun Ji, G. Kunapuli, Jong-Shi Pang.
Model Selection via Bilevel Optimization.
IJCNN '06. Extended Version.

February 18, 2009: Paper presentation by Andrew Goldberg:
M. Karlen, J. Weston, A. Erkan, and R. Collobert.
Large Scale Manifold Transduction.
ICML 2008.

March 4, 2009: Paper presentation by Nate Fillmore:
L. Bottou and O. Bousquet.
The Tradeoffs of Large Scale Learning.
NIPS 2007.

March 11, 2009: Paper presentation by Chris Hinrichs:
L. Xu, K. Crammer, and D. Schuurmans.
Robust Support Vector Machine Training via Convex Outlier Ablation.
AAAI 2006.

April 1,2009: Paper presentation by David Andrzejewski:
Geoffrey E. Hinton.
Training Products of Experts by Minimizing Contrastive Divergence.
Neural Computation 14, 1771-1800 (2002).

April 8, 2009: Paper presentation by Hidayath Ansari:
F. Wang, C. Zhang, T. Li.
Clustering with Local and Global Regularization.
AAAI 2007.

April 22, 2009: Paper presentation by Lisa Torrey:
C. Elkan, K. Noto.
Learning Classifiers from Only Positive and Unlabeled Data.
KDD 2008.

Procedures

As a discussion leader, you should select a paper related to your research, read it carefully if you haven't already, and present it to the group. Focus especially on how the paper relates to or may impact your own research. Leave time for discussion at the end.

As a discussion attendee, you should read the paper before attending so that presenters don't have to teach the whole thing from scratch and can talk about the connections to their research.

Paper presentations are scheduled well ahead of time, usually at the start of the semester by the AIRG organizer. Preliminary research/practice talks can be scheduled as needed by contacting the current organizer. Note that students that have presented a paper have priority when scheduling preliminary research or practice talks.

The current organizer is Ameet Soni (email: soni@cs...)

Tips on choosing papers:

Generally people choose conference-length ones, though longer ones are okay if you only expect people to skim instead of reading in depth.

Don't choose something that only five people in the world will understand.

Choose something that will lead to a discussion that gives you ideas and suggestions for your research.

Previous Semesters

Papers discussed in previous semesters can be found in the archive. See also the schedule for an 838 section from 2004 that covered statistical relational learning.

The AIRG web site is currently maintained by Ameet Soni (CS login: soni).