Least Ambiguous Set-Valued Classifiers With Bounded Error Levels

DOI10.1080/01621459.2017.1395341zbMATH Open1478.62175arXiv1609.00451OpenAlexW2514278201MaRDI QIDQ5229906FDOQ5229906

Larry Wasserman, Jing Lei, Mauricio Sadinle

Publication date: 19 August 2019

Published in: Journal of the American Statistical Association (Search for Journal in Brave)

Abstract: In most classification tasks there are observations that are ambiguous and therefore difficult to correctly label. Set-valued classifiers output sets of plausible labels rather than a single label, thereby giving a more appropriate and informative treatment to the labeling of ambiguous instances. We introduce a framework for multiclass set-valued classification, where the classifiers guarantee user-defined levels of coverage or confidence (the probability that the true label is contained in the set) while minimizing the ambiguity (the expected size of the output). We first derive oracle classifiers assuming the true distribution to be known. We show that the oracle classifiers are obtained from level sets of the functions that define the conditional probability of each class. Then we develop estimators with good asymptotic and finite sample properties. The proposed estimators build on existing single-label classifiers. The optimal classifier can sometimes output the empty set, but we provide two solutions to fix this issue that are suitable for various practical needs.

Full work available at URL: https://arxiv.org/abs/1609.00451

Recommendations

• Efficient set-valued prediction in multi-class classification
• Confidence sets with expected sizes for multiclass classification
• Criteria of efficiency for set-valued classification
• Advances in Intelligent Data Analysis VI

zbMATH Keywords

multiclass classificationreject optionBayes classifierambiguous observationnondeterministic classifieroracle classifier

Mathematics Subject Classification ID

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

Cites Work

• High-dimensional generalized linear models and the lasso
• Title not available (Why is that?)
• Introduction to nonparametric estimation
• Title not available (Why is that?)
• Optimal global rates of convergence for nonparametric regression
• A conformal prediction approach to explore functional data
• Classification with confidence
• Title not available (Why is that?)
• Conditional validity of inductive conformal predictors
• Title not available (Why is that?)
• Distribution-Free Prediction Sets
• Distribution-free Prediction Bands for Non-parametric Regression
• Classification with reject option
• Fast learning rates for plug-in classifiers
• On optimum recognition error and reject tradeoff
• Title not available (Why is that?)
• ML-KNN: A lazy learning approach to multi-label learning
• The uniform convergence of nearest neighbor regression function estimators and their application in optimization
• Learning nondeterministic classifiers
• Classification methods with reject option based on convex risk minimization
• Consistent algorithms for multiclass classification with an abstain option

Cited In (15)

• On soft computing with random fuzzy sets in econometrics and machine learning
• Distribution-Free Prediction Sets for Two-Layer Hierarchical Models
• Three-way decision and conformal prediction: isomorphisms, differences and theoretical properties of cautious learning approaches
• Criteria of efficiency for set-valued classification
• Is distribution-free inference possible for binary regression?
• Optimal rates for nonparametric F-score binary classification via post-processing
• Title not available (Why is that?)
• Uncertainty in lung cancer stage for survival estimation via set-valued classification
• Minimax semi-supervised set-valued approach to multi-class classification
• Agnostic tests can control the type I and type II errors simultaneously
• Nested conformal prediction sets for classification with applications to probation data
• Set-Valued Support Vector Machine with Bounded Error Rates
• A confidence machine for sparse high-order interaction model
• Selective Prediction-Set Models with Coverage Rate Guarantees
• Conformal Prediction: A Gentle Introduction

Uses Software

• UCI-ml
• ML-KNN

This page was built for publication: Least Ambiguous Set-Valued Classifiers With Bounded Error Levels

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5229906)