P747, Q700:  Bayesian and Data Mining
Spring, 2003
"Instructor": Richard M. Shiffrin (contact information below).
Meeting Time, Days, and Room: To Be Arranged

	Recent years have seen marked advances in computing power and speed,
making possible a variety of new techniques for data analysis.  [One
conference that highlights these techniques is NIPS: Neural
Information Processing Systems.]  Many researchers (including the
teacher of the present course, Richard Shiffrin) who use traditional
mathematical modeling and computer simulation techniques are not
conversant with these new methods.  Such researchers and their
students in Cognitive Science and Psychology would gain much from a
course that introduces and surveys some of the most important of these
techniques, and that is the aim of the seminar.  The students and the
instructor will explore together a number of these methods.  Time
constraints will prevent coverage in depth; the aim instead will be to
understand the nature of the techniques, the goals, and possible uses,
by surveying at least one example for each.  One of the goals of this
high level seminar will be the selection of topics (there is not
nearly enough time to cover more than a sampling), the selection
evolving as the course progresses.

	Topics that we will consider include Bayes Nets, Hidden Markov
Models, Kernel Methods, Classification Methods in High Dimensions
(including Bayes Classifiers), Wavelet Methods, Minimum Descriptive
Length for Model Selection, Expection Maximization (EM) Algorithms,
Gibbs Sampling and Bootstrap Methods, Markov Chain Simulation,
Additive Trees, and Support Vector Machines.  A few scaling approaches
may also be considered.

	This will obviously be a high level seminar in which the participants
will all teach each other and learn together.  Enrollment has only the
requirement of permission of instructor.  Richard Shiffrin should be
contacted at shiffrin@Indiana.edu (email), 812-855-4972 (office),
812-336-1041 (home), office: Psychology 350.

	A time and meeting room have not yet been selected and will be
arranged; we will try to do so via email, so contact the instructor as
soon as possible.