|
Activity Number:
|
446
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statistical Computing
|
| Abstract - #307910 |
|
Title:
|
Efficient Large-Margin Semisupervised Learning
|
|
Author(s):
|
Junhui Wang and Xiaotong Shen*+
|
|
Companies:
|
Columbia University and The University of Minnesota
|
|
Address:
|
School of Statistics, Minneapolis, MN, 55455,
|
|
Keywords:
|
Margin ; Semisupservised ; Generalization
|
|
Abstract:
|
In classification, semi-supervised learning involves a large amount of unlabeled data with only a small number of labeled data. This imposes great challenge in that the class probability given input can not be well estimated through labeled data alone. This talk presents a large margin semi-supervised learning method that constructs an efficient loss to measure the contribution of unlabeled instances to classification. An iterative scheme is derived for implementation. The method is examined with two large margin classifiers: support vector machines and psi-learning. Our theoretical and numerical analyses indicate that the method achieves the desired objective of delivering high performance.
|
