A Vector-Contraction Inequality for Rademacher Complexities

DOI10.1007/978-3-319-46379-7_1zbMATH Open1478.68296arXiv1605.00251OpenAlexW2962708723MaRDI QIDQ2830263FDOQ2830263

Authors: Andreas Maurer 

Publication date: 9 November 2016

Published in: Lecture Notes in Computer Science (Search for Journal in Brave)

Abstract: The contraction inequality for Rademacher averages is extended to Lipschitz functions with vector-valued domains, and it is also shown that in the bounding expression the Rademacher variables can be replaced by arbitrary iid symmetric and sub-gaussian variables. Example applications are given for multi-category learning, K-means clustering and learning-to-learn.

Full work available at URL: https://arxiv.org/abs/1605.00251

Recommendations

• 10.1162/153244303321897690
• scientific article; zbMATH DE number 1804106
• Approximation error bounds via Rademacher's complexity
• Rademacher inequalities with applications
• On the $L_p$ norm of the Rademacher projection and related inequalities
• Rademacher Margin Complexity
• The Rademacher Complexity of Linear Transformation Classes
• Estimates of the approximation error using Rademacher complexity: Learning vector-valued functions
• The Khinchin Inequality for Generalized Rademacher Functions
• A \(\Phi \)-entropy contraction inequality for Gaussian vectors

zbMATH Keywords

contraction inequalityRademacher complexities

Mathematics Subject Classification ID

Classification and discrimination; cluster analysis (statistical aspects) (62H30) Learning and adaptive systems in artificial intelligence (68T05) Inequalities; stochastic orderings (60E15) Computational learning theory (68Q32)

Cites Work

• Title not available (Why is that?)
• Transfer bounds for linear feature learning
• Concentration inequalities. A nonasymptotic theory of independence
• Empirical margin distributions and bounding the generalization error of combined classifiers
• Optimal rates for the regularized least-squares algorithm
• 10.1162/153244303321897690
• 10.1162/15324430260185628
• 10.1162/1532443041424300
• Title not available (Why is that?)
• $K$-Dimensional Coding Schemes in Hilbert Spaces
• Regularization techniques for learning with matrices
• On the Performance of Clustering in Hilbert Spaces
• On the best constants in the Khinchin inequality
• The benefit of multitask representation learning

Cited In (14)

• Estimates of the approximation error using Rademacher complexity: Learning vector-valued functions
• Multi-kernel learning for multi-label classification with local Rademacher complexity
• Handling concept drift via model reuse
• Algorithmic Learning Theory
• Weakly Convex Optimization over Stiefel Manifold Using Riemannian Subgradient-Type Methods
• Title not available (Why is that?)
• Robust \(k\)-means clustering for distributions with two moments
• From inexact optimization to learning via gradient concentration
• On strong consistency of kernel \(k\)-means: a Rademacher complexity approach
• Compressive sensing and neural networks from a statistical learning perspective
• Convergence Analysis of Machine Learning Algorithms for the Numerical Solution of Mean Field Control and Games I: The Ergodic Case
• Weight normalized deep neural networks
• Graphical Convergence of Subgradients in Nonconvex Optimization and Learning
• Also for \(k\)-means: more data does not imply better performance

This page was built for publication: A Vector-Contraction Inequality for Rademacher Complexities

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2830263)