PhD Qualifying Examination


"Learning Latent Variable Models: An Overview"

By Mr. Tao Chen


Abstract:

Recently, learning Latent Variable Models(LVMs) has drawn more and more
attentions in the UAI(Uncertainty in AI) community. Unlike traditional
models that describe only relationships among the observed variables, LVMs
allow to include one or more latent or hidden variables. Due to their
stronger representational capability over general Bayesian networks, LVMs
often yield significant improvements to cluster, classify data and to
discover causal relationships from data. In this paper we aim to provide a
literature review on learning LVMs from both theoretical and practical
perspectives. The most commonly used learning approach, the so-called
score-and-search approach, requires a model selection criterion as the
score and a search procedure using the score. Therefore we shall first
discuss various scores and their consistency under the framework of
differential geometry analysis, and then review a number of popular search
algorithms.



Date:     		Friday, 30 January 2004

Time:                   2:00p.m.-4:00p.m.

Venue:                  Room 2302
			lifts 17-18

Committee Members:      Prof. Nevin Zhang (Supervisor)
                        Prof. Albert Chung (Chairperson)
			Prof. Mordecai Golin
			Prof. Dit-Yan Yeung


**** ALL are Welcome ****