MPhil Thesis Defence


"Kernel-based Multiple-Instance Learning"

By

Mr. Pak Ming Cheung


Abstract

In recent years, Multiple-Instance Learning (MIL) problem becomes more and
more popular in machine learning community. Each training object (bag) of
MIL problem is a set of patterns (instances). Label information is only
associated with bags, but not their constituent instances. Moreover, a
positive bag must have at least one positive instance, but may have many
negative instances. Since we can only access the label information of a
bag and a positive bag may be composed of many negative instances, MIL is
more challenging than the traditional supervised learning problem (or
single-instance learning problem). On the other hand, it is fruitful to
study MIL, since many real-world problems, such as drug activity
prediction, are inherently MI problems which cannot be generalized well
under the traditional single-instance learning model. In addition, the
generalization performance of many single-instance learning problems, eg.
Content-based Image Retrieval (CBIR), are found to be improved when they
are casted into an appropriate MIL representation.

In this thesis, I focus on MIL algorithms based on kernel methods, in
particular support vector machines, which have been highly successful in
many machine learning problems. This thesis first discusses how to
re-formulate the SVM to adapt to the MI problem setting by utilizing both
the bag and instance information at the same time. After that, I further
propose how to define a MI kernel over bags, which first assumes the
knowledge of instance labels and then marginalizing this hidden
information out. The resulted bag kernel can then be used in a standard
SVM. I also extend both approaches to the real-valued regression setting,
which is more and more popular in the MIL community. Empirical results
show that the proposed methods have better performance over various
traditional methods.


Date:				Thursday, 24 August 2006

Time:				10:00a.m.-12:00noon

Venue:				Room 5505
				Lifts 25-26

Committee Members:		Dr. James Kwok (Supervisor)
				Dr. Dit-Yan Yeung (Chairperson)
				Dr. Brian Mak


**** ALL are Welcome ****