Aleatoric and epistemic uncertainty in machine learning: an introduction to concepts and methods

DOI10.1007/s10994-021-05946-3OpenAlexW3134774296MaRDI QIDQ2051254

Eyke Hüllermeier, Willem Waegeman

Publication date: 24 November 2021

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

Full work available at URL: https://arxiv.org/abs/1910.09457

zbMATH Keywords

uncertaintyGaussian processesprobabilitycalibrationBayesian inferenceepistemic uncertaintygenerative modelslikelihood-based methodsensemblesconformal predictiondeep neural networksset-valued predictioncredal sets and classifiersversion space learning

Mathematics Subject Classification ID

Learning and adaptive systems in artificial intelligence (68T05)

Related Items (19)

How to measure uncertainty in uncertainty sampling for active learning ⋮ Information efficient learning of complexly structured preferences: elicitation procedures and their application to decision making under uncertainty ⋮ Bayesian-EUCLID: discovering hyperelastic material laws with uncertainties ⋮ Handling epistemic and aleatory uncertainties in probabilistic circuits ⋮ Lymphoma segmentation from 3D PET-CT images using a deep evidential network ⋮ Epistemic uncertainty quantification in deep learning classification by the delta method ⋮ Advanced discretization techniques for hyperelastic physics-augmented neural networks ⋮ Model-independent reconstruction of growth index via Gaussian process ⋮ DEED: deep evidential doctor ⋮ FFNSL: Feed-forward neural-symbolic learner ⋮ Uncertainty quantification in scientific machine learning: methods, metrics, and comparisons ⋮ Responsible model deployment via model-agnostic uncertainty learning ⋮ Three-way decision and conformal prediction: isomorphisms, differences and theoretical properties of cautious learning approaches ⋮ Neural-physics multi-fidelity model with active learning and uncertainty quantification for GPU-enabled microfluidic concentration gradient generator design ⋮ Quality measures for the evaluation of machine learning architectures on the quantification of epistemic and aleatoric uncertainties in complex dynamical systems ⋮ Some models are useful, but how do we know which ones? Towards a unified Bayesian model taxonomy ⋮ Finite electro-elasticity with physics-augmented neural networks ⋮ Assessments of epistemic uncertainty using Gaussian stochastic weight averaging for fluid-flow regression ⋮ Multifidelity data fusion in convolutional encoder/decoder networks

Uses Software

• IsolationForest
• EfficientNet

Cites Work

• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Reliable classification: learning classifiers that distinguish aleatoric and epistemic uncertainty
• Support vector data description
• Learning valued preference structures for solving classification problems
• Axiomatic characterizations of information measures
• Comparing probability measures using possibility theory: a notion of relative peakedness
• Possibility theory and statistical reasoning
• On the uniqueness of possibilistic measure of uncertainty and information
• Default reasoning and possibility theory
• On random sets and belief functions
• The transferable belief model
• A semantics for possibility theory based on likelihoods
• Credal networks
• The naive credal classifier
• Consistent algorithms for multiclass classification with an abstain option
• Prediction algorithms and confidence measures based on algorithmic randomness theory.
• On the order of triangular norms -- comments on ``A triangular norm hierarchy by E. Cretu
• An introduction to variational methods for graphical models
• Evaluating credal classifiers by utility-discounted predictive accuracy
• Unifying practical uncertainty representations. I. Generalized \(p\)-boxes
• An introduction to the imprecise Dirichlet model for multinomial data
• Likelihood-based belief function: justification and some extensions to low-quality data
• Regression conformal prediction with random forests
• Optimal Uncertainty Quantification
• Probability estimation with machine learning methods for dichotomous and multicategory outcome: Theory
• A NON-SPECIFICITY MEASURE FOR CONVEX SETS OF PROBABILITY DISTRIBUTIONS
• Disaggregated total uncertainty measure for credal sets
• Belief functions and statistical inference
• A new metric for probability distributions
• ENTROPY AND SPECIFICITY IN A MATHEMATICAL THEORY OF EVIDENCE
• Science from Fisher Information
• Optimal Design of Experiments
• Classification with reject option
• Strictly Proper Scoring Rules, Prediction, and Estimation
• On optimum recognition error and reject tradeoff
• The Nearest Neighbor Classification Rule with a Reject Option
• An invariant form for the prior probability in estimation problems
• Random forests

This page was built for publication: Aleatoric and epistemic uncertainty in machine learning: an introduction to concepts and methods