Learning using privileged information: SVM+ and weighted SVM

DOI10.1016/J.NEUNET.2014.02.002MaRDI QIDQ2339396zbMATH OpenOpenAlexDBLPWikidataFDO

Authors Maksim Lapin, Matthias Hein, Bernt Schiele

Publication date 1 April 2015

Published in Neural Networks (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1306.3161

zbMATH Keywords

importance weightingprior knowledgeprivileged informationSVMSVM+weighted SVM

Mathematics Subject Classification ID

Learning and adaptive systems in artificial intelligence (68T05)

Abstract: Prior knowledge can be used to improve predictive performance of learning algorithms or reduce the amount of data required for training. The same goal is pursued within the learning using privileged information paradigm which was recently introduced by Vapnik et al. and is aimed at utilizing additional information available only at training time -- a framework implemented by SVM+. We relate the privileged information to importance weighting and show that the prior knowledge expressible with privileged features can also be encoded by weights associated with every training example. We show that a weighted SVM can always replicate an SVM+ solution, while the converse is not true and we construct a counterexample highlighting the limitations of SVM+. Finally, we touch on the problem of choosing weights for weighted SVMs when privileged features are not available.

Recommendations

• Robust learning with imperfect privileged information
• Learning using privileged information: similarity control and knowledge transfer
• A new learning paradigm: learning using privileged information
• Improved multi-view privileged support vector machine
• Linear feature-weighted support vector machine

Cites work

• A new learning paradigm: learning using privileged information
• Choosing multiple parameters for support vector machines
• Confidence-weighted linear classification for text categorization
• Convex analysis and nonlinear optimization. Theory and examples
• Estimation of Dependences Based on Empirical Data
• scientific article; zbMATH DE number 1804115 (Why is no real title available?)
• scientific article; zbMATH DE number 1931841 (Why is no real title available?)
• scientific article; zbMATH DE number 2107836 (Why is no real title available?)
• scientific article; zbMATH DE number 5060482 (Why is no real title available?)
• Improving predictive inference under covariate shift by weighting the log-likelihood function
• Input-dependent estimation of generalization error under covariate shift
• Predictive learning with structured (grouped) data
• Sample Selection Bias as a Specification Error
• Training a Support Vector Machine in the Primal

Cited in

(16)

• Improved multi-view privileged support vector machine
• KOC+: kernel ridge regression based one-class classification using privileged information
• Instance weighting through data imprecisiation
• Robust learning with imperfect privileged information
• A Riemannian gossip approach to subspace learning on Grassmann manifold
• A new learning paradigm: learning using privileged information
• An easy-to-hard learning paradigm for multiple classes and multiple labels
• scientific article; zbMATH DE number 7255066 (Why is no real title available?)
• A method to enrich experimental datasets by means of numerical simulations in view of classification tasks
• scientific article; zbMATH DE number 7625168 (Why is no real title available?)
• Dual iterative hard thresholding
• The benefits of modeling slack variables in SVMs
• Learning using privileged information: similarity control and knowledge transfer
• Twin support vector machines with privileged information
• AdaBoost-based transfer learning with privileged information
• A novel stochastic configuration network with iterative learning using privileged information and its application

Describes a project that uses

Uses Software

• UCI
• LIBSVM

This page was built for publication: Learning using privileged information: SVM+ and weighted SVM

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2339396)