PLACE: 319 Snedecor

SPEAKER:
Yuguo Chen
Department of Statistics, Stanford University

TITLE:
Conditional Inference on Zero-One Tables:  A Sequential Importance
Sampling Approach

ABSTRACT:

The Monte Carlo method of sequential importance sampling (SIS) has been shown
to be a versatile and powerful tool for solving complex problems in dynamic
systems.  We describe a sequential importance sampling approach to making
conditional inferences about zero-one tables, a problem which is not inherently
dynamic.  Our procedure compares favorably with Markov Chain Monte Carlo
techniques.  We apply our method to test ecological theories about competition
between species in Darwin's finch data.  We discuss the insights that our
approach to this problem provides for developing an efficient SIS methodology.
We briefly describe other general principles behind efficient SIS algorithms we
have developed for inference on genealogical trees, permutation tests on
truncated data and filtering and smoothing in hidden Markov models.
 
 

COFFEE: 3:45 p.m., 104 Snedecor