University of Southern California
Ray R. Irani Hall
Molecular and Computational Biology
Computational Biology Colloquium
Fei Sha
Department of Computer Science
USC
“How to harvest information from high dimension data with statistical learning techniques”
Abstract:
Technologies for sensing and acquiring data have been significantly advanced. For example, a large number of inexpensive sensors can be readily deployed, measuring a vast number of physical quantities in many modalities over a long period of time. Similarly, in other application domains, such as bioinformatics and genomics, tremendous amount of data can be and have been amassed.
A fundamental challenge remains open, though. How can we reason with, extract information and mine knowledge from such data? In particular, these data arise often in high dimensionality which severely impedes our intuition and ability for modeling, visualizing and analyzingthem.
Statistical machine learning techniques have been very effective in addressing these challenges, by reducing dimensionality and revealing intrinsic structures that underpin high dimensional data. Consequently, these techniques have become essential and indispensable in modern data analysis. In this talk, I will present a
few examples of such techniques, including some that were developed by me and my collaborators. We applied those techniques to images, text documents and data sets of protein functions. We have gained many useful insights about those data through our analysis.
Thursday, April 29, 2010
2:00 pm
RRI 101
