Support Vector Machine algorithm for regression and classification
The software is an implementation of the Support Vector Machine (SVM) algorithm that was invented and developed by Vladimir Vapnik and his co-workers at AT&T Bell Laboratories. The specific implementation reported here is an Active Set method for solving a quadratic optimization problem that forms the major part of any SVM program. The implementation is tuned to specific constraints generated in the SVM learning. Thus, it is more efficient than general-purpose quadratic optimization programs. A decomposition method has been implemented in the software that enables processing large data sets. The size of the learning data is virtually unlimited by the capacity of the computer physical memory. The software is flexible and extensible. Two upper bounds are implemented to regulate the SVM learning for classification, which allow users to adjust the false positive and false negative rates. The software can be used either as a standalone, general-purpose SVM regression or classification program, or be embedded into a larger software system.
- Short Name / Acronym:
- ActiveSVM; 001690MLTPL00
- Version:
- 00
- Programming Language(s):
- Medium: X; OS: Microsoft Windows Systems (Win 95 and up, NT, XP), UNIX and LINUX.; Compatibility: Multiplatform
- Research Organization:
- Argonne National Laboratory (ANL), Argonne, IL (United States)
- Sponsoring Organization:
- USDOE
- DOE Contract Number:
- W31-109-31eng
- OSTI ID:
- 1230694
- Country of Origin:
- United States
Similar Records
SUPPORT VECTOR MACHINES FOR BROAD AREA FEATURE CLASSIFICATION IN REMOTELY SENSED IMAGES
Classification Using Support Vector Machines with Uncertainty Quantification