Label noise is also an important challenge in classification. Indeed, experts often make errors when they label instances, which can hurt the performances of inferred models. In this context, I wrote a thorough survey of the label noise literature. I also adapted an approach developed by Lawrence and Schölkopf to handle label noise in the context of automated electrocardiogram segmentation with hidden Markov models. Mutual information estimation in the presence of label noise is considered in another of my publications.
I have been chair/organiser for the special session “Label noise in Classification” at the ESANN 2014 conference and editor for the special issue “Advances in Learning with Label Noise” in the Neurocomputing journal. See my CFPs for more details and other topics.
- B. Frénay, A. Kaban. Special issue on advances in learning with label noise. Neurocomputing, 160, 2015, p. 1-2.
- Frénay, B., Verleysen, M. Classification in the Presence of Label Noise: a Survey. IEEE Trans. Neural Networks and Learning Systems, 25(5), 2014, p. 845-869.
- Frénay, B., Kabán A. A Comprehensive Introduction to Label Noise. In Proc. ESANN, Bruges, Belgium, 23-25 April 2014, p. 667–676.
- Frénay, B., Verleysen, M. Pointwise Probability Reinforcements for Robust Statistical Inference. Neural networks, , 50, 124-141, 2014.
- Frénay, B., Doquire, G., Verleysen, M. Estimating mutual information for feature selection in the presence of label noise. Computational Statistics & Data Analysis, 71, 832-848, 2014.
- Frénay, B., de Lannoy, G., Verleysen, M. Label Noise-Tolerant Hidden Markov Models for Segmentation: Application to ECGs. In Proc. ECML-PKDD 2011, p. 455-470.