A novel parameter refinement approach to one class support vector machine

Authors:
Trung Le;Dat Tran;Wanli Ma;Dharmendra Sharma
Affiliations:
Faculty of Information Sciences and Engineering, University of Canberra, ACT, Australia;Faculty of Information Sciences and Engineering, University of Canberra, ACT, Australia;Faculty of Information Sciences and Engineering, University of Canberra, ACT, Australia;Faculty of Information Sciences and Engineering, University of Canberra, ACT, Australia
Venue:
ICONIP'11 Proceedings of the 18th international conference on Neural Information Processing - Volume Part II
Year:
2011

Citing 11
Cited 0

A probabilistic resource allocating network for novelty detection

Neural Computation
Outliers in statistical pattern recognition and an application to automatic chromosome classification

Pattern Recognition Letters
Pattern Recognition and Neural Networks

Pattern Recognition and Neural Networks
Support Vector Machines for Classification in Nonstandard Situations

Machine Learning
Local Expert Autoassociators for Anomaly Detection

ICML '00 Proceedings of the Seventeenth International Conference on Machine Learning
A decision-theoretic generalization of on-line learning and an application to boosting

EuroCOLT '95 Proceedings of the Second European Conference on Computational Learning Theory
Support Vector Data Description

Machine Learning
Learning Minimum Volume Sets

The Journal of Machine Learning Research
Consistency and Convergence Rates of One-Class SVMs and Related Algorithms

The Journal of Machine Learning Research
Review:

Neural Computation
A Small Sphere and Large Margin Approach for Novelty Detection Using Training Data with Outliers

IEEE Transactions on Pattern Analysis and Machine Intelligence

Quantified Score

Hi-index	0.00

Visualization

Abstract

One-Class Support Vector Machine employs a grid parameter selection process to discover the best parameters for a given data set. It is assumed that two separate trade-off parameters are assigned to normal and abnormal data samples, respectively. However, this assumption is not always true because data samples have different contributions to the construction of hypersphere or hyperplane decision boundary. In this paper, we introduce a new iterative learning process that is carried out right after the grid parameter selection process to refine the trade-off parameter value for each sample. In this learning process, a weight is assigned to each sample to represent the contribution of that sample and is iteratively refined. Experimental results performed on a number of data sets show a better performance for the proposed approach.