Peak criterion for choosing Gaussian kernel bandwidth in Support Vector Data Description
Authors:
- 1 June 2017
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 32-39
Abstract
Support Vector Data Description (SVDD) is a machine learning technique used for single class classification and outlier detection. SVDD formulation with kernel function provides a flexible boundary around data. The value of kernel function parameters affects the nature of data boundary. For example, it is observed that with Gaussian kernel, as the value of kernel bandwidth is lowered, the data boundary changes from spherical to wiggly. The spherical data boundary leads to underfitting and extremely wiggly data boundary leads to overfitting. In this paper we propose an empirical criteria to obtain a good value of Gaussian kernel bandwidth which provides a smooth boundary capturing the essential visual features of the data.Keywords
All Related Versions
This publication has 15 references indexed in Scilit:
- Two methods of selecting Gaussian kernel parameters for one-class SVM and their application to fault detectionKnowledge-Based Systems, 2014
- Just‐in‐time reorganized PCA integrated with SVDD for chemical process monitoringAIChE Journal, 2013
- SVDD-based outlier detection on uncertain dataKnowledge and Information Systems, 2012
- One-class classification-based control charts for multivariate process monitoringIIE Transactions, 2009
- Support vector machine in machine condition monitoring and fault diagnosisMechanical Systems and Signal Processing, 2007
- SVDD-Based Pattern DenoisingNeural Computation, 2007
- One-Class Classification for Mapping a Specific Land-Cover Class: SVDD Classification of FenlandIEEE Transactions on Geoscience and Remote Sensing, 2007
- Some Properties of the Gaussian Kernel for One Class LearningLecture Notes in Computer Science, 2007
- Support Vector Data DescriptionMachine Learning, 2004
- A kernel-distance-based multivariate control chart using support vector methodsInternational Journal of Production Research, 2003