|
|
 |
 News
Colloquia
Identification of a small number of highly predictive genes for disease status
Dr. Md. Rafiul Hassan
University of Melbourne, Australia
May 14, 2009
JEC 3117, 4:00 p.m. to 5:00 p.m.
Refreshments at 3:30 p.m.
Abstract:
Current micro-array data can have dimensions up to approximately 20,000.
It is anticipated that this dimension could reach up to one million (SNP
Array 6.0 has more than 946,000 probes for the detection of copy number
variation). This, in turn, shows that the capacity of micro-arrays is
growing into the hundreds of thousands of dimensions. The large
dimension of the data introduces the problem of the ‘curse of
dimensionality’, which, in turn, implies that an exponential number of
instances are needed to model the distribution (e.g. for a dataset with
100 dimensions with each feature containing two values, the required
data instances are of the order of 2100 in order to represent the
underlying distribution, which is highly impractical). Fortunately, some
of these features are independent and many do not contribute in
discriminating the dataset. In this talk we describe robust feature
selection techniques based on area under the receiver operating
characteristics curve to effectively classify the disease status using
gene expression data.
Bio:
Dr. Md. Rafiul Hassan is working as a research fellow at the Department
of Computer Science and Software Engineering in the University of
Melbourne, Australia. He received a PhD in 2007 from the University of
Melbourne and a B.Sc (Engg.) in Electronics and Computer Science in 2000
from Shah Jalal University of Science and Technology, Bangladesh. His
research interests include data mining, support vector machine, feature
selection and Receiver Operating Characteristics Curve (ROC) with a
particular focus on developing machine learning tools for classifying
gene expression data. Before joining the University of Melbourne, Dr.
Hassan worked as an Assistant Professor at the Department of Computer
Science and Engineering in Shah Jalal University of Science and
Technology, Bangladesh. He is currently involved in research and
development projects for effective classification of Bioinformatics
data. He is the author of around 20 papers in recognized international
journals and conferences. He is a member of Australian Society of
Operations Research (ASOR), IEEE and IEEE Computer society; and is
involved in several Program Committees of international conferences. He
also serves as the reviewer of a few renowned journals such as BMC
Cancer, Information Sciences, Digital Signal Processing, and Computer
Communications.
Hosted by: Dr. Mohammed Zaki (x6340)
Last updated: May 1, 2009
|
|
|
|
