Error Control for Support Vector Machines

منو

" Error Control for Support Vector Machines "

Document Type	:	Latin Dissertation
Language of Document	:	English
Record Number	:	148746
Doc. No	:	ET20538
Main Entry	:	Mark A. Davenport
Title Proper	:	Error Control for Support Vector Machines
Note	:	This document is digital این مدرک بصورت الکترونیکی می باشد
Abstract	:	In binary classification there are two types of errors, and in many applicationsthese may have very different costs. We consider two learning frameworks that ad-dress this issue: minimax classification, where we seek to minimize the maximumof the false alarm and miss rates, and Neyman-Pearson (NP) classification, wherewe seek to minimize the miss rate while ensuring the false alarm rate is less than aspecified level a. We show that our approach, based on cost-sensitive support vectormachines, significantly outperforms methods typically used in practice. Our resultsalso illustrate the importance of heuristics for improving the accuracy of error rateestimation in this setting. We then reduce anomaly detection to NP classificationby considering a second class of points, allowing us to estimate minimum volumesets using algorithms for NP classification. Comparing this approach with traditionalone-class methods, we find that our approach has several advantages..
Subject	:	Electericl tess
	:	برق
electronic file name	:	TL43675.pdf
Title and statement of responsibility and	:	Error Control for Support Vector Machines [Thesis]