|
Activity Number:
|
335
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Nonparametric Statistics
|
| Abstract - #306889 |
|
Title:
|
Finding an Approximate Solution Path of Support Vector Machines for Large Datasets
|
|
Author(s):
|
Zhenhuan Cui*+ and Yoonkyung Lee
|
|
Companies:
|
The Ohio State University and The Ohio State University
|
|
Address:
|
1958 Neil Avenue, Columbus, 43210,
|
|
Keywords:
|
solution path ; support vector machine ; algorithm ; classification ; regularization ; large datasets
|
|
Abstract:
|
The solution path of support vector machines (SVM) contains the entire set of solutions at every value of the regularization parameter that controls the complexity of a fitted model. The algorithm for the solution path has recently been extended from binary cases to multi-category cases. This algorithm greatly facilitates the computation of SVM by sequentially constructing the whole spectrum of solutions. However, large datasets and the choice of a flexible kernel may pose a computational challenge to the sequentially updating algorithm. In this paper, we borrow the idea of basis thinning to alleviate the computational load for large datasets and propose a method for approximate solution paths. In addition, some related computational issues are discussed and the effectiveness of the algorithm is demonstrated for some benchmark datasets.
|
