Login
Computer Science @ Rochester
Monday, April 28, 2003
11:00 AM
CSB 209
Ph.D. Thesis Proposal
Tao Li
University of Rochester
Knowledge Discovery from Labeled and Unlabeled Data
Knowledge discovery, also known as data mining, is the process of automatic extraction of novel, useful and understandable patterns/models from large datasets. Although data mining as a discipline has matured considerably and there exists a multitude of scalable algorithms that transform oceans of bits in very large databases into interpretable patterns and predictive models, much work still needs to be done for efficiently and accurately learning in a wide variety of scenarios.

The goal of the thesis is to develop algorithms for learning from labeled and unlabeled data. We first propose a simple and efficient multi-class classification approach via generalized discriminant analysis. Second, we study the methods for automatically generating hierarchical structures to facilitate classification. Third, we introduce the concept of co-updating, a semi-supervised learning algorithm from different information sources. Finally, we summarize our previous work that is closely related to the proposal and discuss future research.