Making Risk Minimization Tolerant to Label Noise

  title={Making Risk Minimization Tolerant to Label Noise},
  author={Aritra Ghosh and Naresh Manwani and P. S. Sastry},
In many applications, the training data, from which one need s to learn a classifier, is corrupted with label noise. Many st andard algorithms such as SVM perform poorly in presence of label no ise. In this paper we investigate the robustness of risk mini ization to label noise. We prove a sufficient condition on a loss funct io for the risk minimization under that loss to be tolerant t o uniform label noise. We show that the 0 − 1 loss, sigmoid loss, ramp loss and probit loss satisfy this c… CONTINUE READING
Highly Cited
This paper has 52 citations. REVIEW CITATIONS
Recent Discussions
This paper has been referenced on Twitter 2 times over the past 90 days. VIEW TWEETS


Publications citing this paper.
Showing 1-10 of 31 extracted citations

53 Citations

Citations per Year
Semantic Scholar estimates that this publication has 53 citations based on the available data.

See our FAQ for additional information.


Publications referenced by this paper.
Showing 1-10 of 45 references

A Probabilistic Theory o f Pattern Recognition

  • L. Devroye, L. Gyorfi, G. Lugosi
  • Springer-Verlag, New York
  • 1996
Highly Influential
4 Excerpts


  • C. Scott, G. Blanchard
  • Handy, Classification with as ymmetric label…
  • 2013
2 Excerpts

UCI machine learning repository

  • K. Bache, M. Lichman
  • 2013
1 Excerpt

Functional gradient ascent for pr bit regression

  • S. Zheng
  • Pattern Recogn .
  • 2012

Similar Papers

Loading similar papers…