|
Activity Number:
|
251
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistical Computing
|
| Abstract - #309886 |
|
Title:
|
Feature Selection for Large Data
|
|
Author(s):
|
Peng Liu*+ and Jiayang Sun
|
|
Companies:
|
Case Western Reserve University and Case Western Reserve University
|
|
Address:
|
26241 Lake Shore Blvd, Euclid, OH, 44132,
|
|
Keywords:
|
data mining ; large data ; feature selection ; mixture distribution ; partial EM ; intrusion detection
|
|
Abstract:
|
A typical challenge in data mining is that data can be too large to be loaded into a computer program once and for all for an analysis, or data come sequentially in streams, so it is necessary to work on pieces of the data and then combine the information from different pieces to obtain the whole picture. In this talk we describe our recent research in developing techniques for feature selection and mixture estimation for large data. Their performance is evaluated by asymptotic analysis and simulation, and compared with standard algorithms. The application of our proposed methods in intrusion detection is demonstrated on the KDD Cup 1999 dataset. (Part of the talk is based on joint work with J. Chen and Z. Zhang.)
|
