Progressive refinement for support vector machines

  • Kiri L. Wagstaff
  • Michael Kocurek
  • Dominic Mazzoni
  • Benyang Tang
Article

DOI: 10.1007/s10618-009-0149-y

Cite this article as:
Wagstaff, K.L., Kocurek, M., Mazzoni, D. et al. Data Min Knowl Disc (2010) 20: 53. doi:10.1007/s10618-009-0149-y

Abstract

Support vector machines (SVMs) have good accuracy and generalization properties, but they tend to be slow to classify new examples. In contrast to previous work that aims to reduce the time required to fully classify all examples, we present a method that provides the best-possible classification given a specific amount of computational time. We construct two SVMs: a “full” SVM that is optimized for high accuracy, and an approximation SVM (via reduced-set or subset methods) that provides extremely fast, but less accurate, classifications. We apply the approximate SVM to the full data set, estimate the posterior probability that each classification is correct, and then use the full SVM to reclassify items in order of their likelihood of misclassification. Our experimental results show that this method rapidly achieves high accuracy, by selectively devoting resources (reclassification) only where needed. It also provides the first such progressive SVM solution that can be applied to multiclass problems.

Keywords

Support vector machines Efficiency Reclassification 

Copyright information

© The Author(s) 2009

Authors and Affiliations

  • Kiri L. Wagstaff
    • 1
  • Michael Kocurek
    • 2
  • Dominic Mazzoni
    • 1
    • 3
  • Benyang Tang
    • 1
  1. 1.Jet Propulsion LaboratoryCalifornia Institute of TechnologyPasadenaUSA
  2. 2.California Institute of TechnologyPasadenaUSA
  3. 3.Google Inc.Santa MonicaUSA

Personalised recommendations