Advanced Research Topics in AI CS 591/691 Section 002 (1 hr) (also available not-for-credit)

George Luger, FEC 325, 277-3204, luger@cs.unm.edu
Andrew P. Kosoresow, FEC 329, 277-8432, kos@cs.unm.edu
Carl Stern, FEC 327, 277-9140, stern@cs.unm.edu

TIME: 11:30am-12:30pm
PLACE: FEC 141

This year we plan to focus on algorithms for machine learning. We will start by surveying traditional approaches, such as parameter setting, classifier systems, versions spaces, explanation based learning, and case based reasoning. We will then examine various theoretical results on learning and try out a variety of learning algorithms, including packages MLC++ and the Luger and Stubblefield software on data sets such as the one developed at UC Irvine. Time allowing, we will address current topics, including Hidden Markov Models, PAC learning, and extensions of the ID3 algorithm, such as C4.5, "bagging," and "boosting." Seminars will feature presentations by visitors and surveys by the seminar leaders. Student participation (such as presenting a paper or reporting experimental results) is encouraged.

For admission on a not-for-credit basis, you must read the material and enter discussions. For admission for credit, the presentation of a paper and active discussion of results is required.