We introduce a computational design for pattern detection based on a tree-structured network of support vector machines (SVMs). An SVM is associated with each cell in a recursive ...
Abstract--Practical applications call for efficient model selection criteria for multiclass support vector machine (SVM) classification. To solve this problem, this paper develops ...
Several theoretical methods have been developed in the past years to evaluate the generalization ability of a classifier: they provide extremely useful insights on the learning ph...
: Matrix-pattern-oriented Least Squares Support Vector Classifier (MatLSSVC) can directly classify matrix patterns and has a superior classification performance than its vector ver...
Background modeling is an essential and important part of many high-level video processing applications. Recently, the Support Vector Data Description (SVDD) has been introduced f...
Alireza Tavakkoli, Mircea Nicolescu, George Bebis,...