Random classification noise defeats all convex potential boosters

DOI10.1007/s10994-009-5165-zzbMath1470.68139WikidataQ56114422 ScholiaQ56114422MaRDI QIDQ1959553

Rocco A. Servedio, Philip M. Long

Publication date: 7 October 2010

Published in: Machine Learning (Search for Journal in Brave)

Full work available at URL: https://doi.org/10.1007/s10994-009-5165-z

zbMATH Keywords

learning theory; boosting; convex loss; noise-tolerant learning; misclassification noise; potential boosting

Mathematics Subject Classification ID

62H30: Classification and discrimination; cluster analysis (statistical aspects)

68T05: Learning and adaptive systems in artificial intelligence

Related Items

Unnamed Item, Surprising properties of dropout in deep networks, Unnamed Item, Boosting in the Presence of Outliers: Adaptive Classification With Nonconvex Loss Functions, A Framework of Learning Through Empirical Gain Maximization, Robust Support Vector Machines for Classification with Nonconvex and Smooth Losses, Classification with asymmetric label noise: consistency and maximal denoising, The risk of trivial solutions in bipartite top ranking, Soft-max boosting, Binary classification with corrupted labels, On the noise estimation statistics, Robustness of learning algorithms using hinge loss with outlier indicators, A non-intrusive correction algorithm for classification problems with corrupted data, Robust Algorithms via PAC-Bayes and Laplace Distributions

Uses Software

• AdaBoost.MH

Cites Work

• Unnamed Item
• A decision-theoretic generalization of on-line learning and an application to boosting
• A geometric approach to leveraging weak learners
• Additive logistic regression: a statistical view of boosting. (With discussion and a rejoinder by the authors)
• Population theory for boosting ensembles.
• On the Bayes-risk consistency of regularized boosting methods.
• Statistical behavior and consistency of classification methods based on convex risk minimization.
• Boosting a weak learning algorithm by majority
• Improved boosting algorithms using confidence-rated predictions
• Boosting with early stopping: convergence and consistency
• 10.1162/153244304773936072
• 10.1162/153244304773936108
• Learning Theory
• Boosting in the presence of noise
• Soft margins for AdaBoost
• An adaptive version of the boost by majority algorithm