Abstract
Receiver operating characteristic (ROC) curves are a plot of a ranking classifier's true-positive rate versus its false-positive rate, as one varies the threshold between positive and negative classifications across the continuum. The area under the ROC curve offer a measure of the discriminatory power of machine learning algorithms that is independent of class distribution, via its equivalence to Mann-Whitney U-statistics. This measure has recently been extended to cover problems of discriminating three and more classes. In this case, the area under the curve generalizes to the volume under the ROC surface. In this paper, we show how a multi-class classifier can be trained by directly maximizing the volume under the ROC surface. This is accomplished by first approximating the discrete U-statistic that is equivalent to the volume under the surface in a continuous manner, and then maximizing this approximation by gradient ascent.
Original language | English |
---|---|
Title of host publication | Computer Aided Systems Theory - EUROCAST 2007 - 11th International Conference on Computer Aided Systems Theory, Revised Selected Papers |
Publisher | IUCTC Las Palmas de Gran Canaria |
Pages | 878-885 |
Number of pages | 8 |
ISBN (Print) | 9783540758662 |
DOIs | |
Publication status | Published - 2007 |
Event | International Conference Computer Aided Systems Theory EUROCAST 2007 - Las Palmas, Spain Duration: 12 Feb 2007 → 17 Feb 2007 http://www.ulpgc.es |
Publication series
Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
---|---|
Volume | 4739 LNCS |
ISSN (Print) | 0302-9743 |
ISSN (Electronic) | 1611-3349 |
Conference
Conference | International Conference Computer Aided Systems Theory EUROCAST 2007 |
---|---|
Country/Territory | Spain |
City | Las Palmas |
Period | 12.02.2007 → 17.02.2007 |
Internet address |
Keywords
- Discrimination analysis
- Machine learning
- Multi-class ROC analysis