ALERT: This system is being upgraded on Tuesday December 12. It will not be available for use for several hours that day while the upgrade is in progress. Deposits to DukeSpace will be disabled on Monday December 11, so no new items are to be added to the repository while the upgrade is in progress. Everything should be back to normal by the end of day, December 12.

Show simple item record

Towards Better Representations with Deep/Bayesian Learning

dc.contributor.advisor Carin, Lawrence Li, Chunyuan 2019-04-02T16:26:43Z 2019-04-02T16:26:43Z 2018
dc.description.abstract <p>Deep learning and Bayesian Learning are two popular research topics in machine learning. They provide the flexible representations in the complementary manner. Therefore, it is desirable to take the best from both fields. This thesis focuses on the intersection of the two topics— enriching one with each other. Two new research topics are inspired: Bayesian deep learning and Deep Bayesian learning.</p><p>In Bayesian deep learning, scalable Bayesian methods are proposed to learn the weight uncertainty of deep neural networks (DNNs). On this topic, I propose the preconditioned stochastic gradient MCMC methods, then show its connection to Dropout, and its applications to modern network architectures in computer vision and natural language processing. </p><p>In Deep Bayesian learning: DNNs are employed as powerful representations of conditionals in traditional Bayesian models. I will focus on understanding the recent adversarial learning methods for joint distribution matching, through which several recent bivariate adversarial models are unified. It further raises the non-identifiability issues in bidirectional adversarial learning, and propose ALICE algorithms: a conditional entropy framework to remedy the issues. The derived algorithms show significant improvement in the tasks of image generation and translation, by solving the non-identifiability issues.</p>
dc.subject Artificial intelligence
dc.subject Statistics
dc.subject Computer science
dc.subject adversarial learning
dc.subject Bayesian learning
dc.subject deep learning
dc.subject generative models
dc.subject neural networks
dc.title Towards Better Representations with Deep/Bayesian Learning
dc.type Dissertation
dc.department Electrical and Computer Engineering

Files in this item


This item appears in the following Collection(s)

Show simple item record