BEGIN:VCALENDAR
METHOD:PUBLISH
PRODID:-//Apple Computer\, Inc//iCal 1.0//EN
X-WR-CALNAME;VALUE=TEXT:USC
VERSION:2.0
BEGIN:VEVENT
DESCRIPTION:\n
PhD Candidate: Rob Brekelmans\n
\n
Title: Information Geometry of Annealing Paths for Inference and Estimation\n
\n
Date: Wednesday, May 18, 2022, 3:00 PM PST\n
\n
Location: Seeley G Mudd (SGM) Room 226 or https://usc.zoom.us/j/91562244468\n
\n
Committee: Greg Ver Steeg, Aram Galstyan, Aiichiro Nakano, Assad Oberai\n
\n
Estimating normalization constants and (log) marginal likelihoods is a fundamental problem in probabilistic machine learning, playing a role in maximum likelihood learning, variational inference, and estimation of information theoretic quantities. Importance sampling, where samples from a tractable proposal distribution are reweighted based on their probability under the target density, is at the heart of many successful solutions including the evidence lower bound (ELBO), importance weighted-autoencoder (IWAE), and annealed importance sampling (AIS).\n
\n
In this thesis, we provide unifying perspectives on these methods and propose methodological improvements. We propose a general approach for deriving extended state-space importance sampling bounds, leading to novel AIS and energy-based methods which can accurately estimate large values of mutual information. We consider extended state space bounds in the context of variational inference, and finally propose a new one-parameter family of annealing paths which generalize the ubiquitous geometric averaging path and can improve estimation performance on example tasks. \n
SEQUENCE:5
DTSTART:20220518T150000
LOCATION:SGM 226
DTSTAMP:20220518T150000
SUMMARY:PhD Defense - Rob Brekelmans
UID:EC9439B1-FF65-11D6-9973-003065F99D04
DTEND:20220518T170000
END:VEVENT
END:VCALENDAR