The nature of statistical learning theory
The nature of statistical learning theory
Machine Learning
Machine Learning
Support Vector Data Description
Machine Learning
Kernel Methods for Pattern Analysis
Kernel Methods for Pattern Analysis
Monocular Pedestrian Detection: Survey and Experiments
IEEE Transactions on Pattern Analysis and Machine Intelligence
Mercer’s theorem, feature maps, and smoothing
COLT'06 Proceedings of the 19th annual conference on Learning Theory
Unsupervised change-detection in retinal images by a multiple-classifier approach
MCS'10 Proceedings of the 9th international conference on Multiple Classifier Systems
| Hi-index | 0.10 |
Change detection is of great interest in many applications where image processing is involved; it is an essential step of image analysis in various applications, including remote sensing (e.g., assessing changes in a forest ecosystems), surveillance (e.g., monitoring ice surface changes to detect glacier flows and detecting changes caused by forest defoliation), and medical diagnosis (e.g., monitoring tumor growth). This paper aims to classify changes in multi-acquisition data using kernel-based support vector data description (SVDD). SVDD is a well-known method that enables one to map the data into a high-dimensional feature space in which a hypersphere encloses most patterns belonging to the unchanged class. In this work, we propose a new kernel function that combines the characteristics of basic kernel functions with new information about the feature distribution and the dependencies among samples. The dependencies among samples are characterized using copula theory; to our knowledge, this is the first time copula theory has been used in the SVDD framework. We demonstrate that the proposed kernel function is robust and has higher performance compared with classical support vector machine (SVM) and support vector data description (SVDD) methods.