Confusion Matrices: A Unified Theory

The confusion matrix is a key tool for understanding and evaluating models in supervised classification problems. Various matrices are proposed depending on the problem framework: single-label, multi-label, or even soft-label restricted to probability distributions. However, most of these approaches...

Full description

Saved in:
Bibliographic Details
Main Authors: Johan Erbani, Pierre-Edouard Portier, Elod Egyed-Zsigmond, Diana Nurbakova
Format: Article
Language:English
Published: IEEE 2024-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10769075/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850259323537588224
author Johan Erbani
Pierre-Edouard Portier
Elod Egyed-Zsigmond
Diana Nurbakova
author_facet Johan Erbani
Pierre-Edouard Portier
Elod Egyed-Zsigmond
Diana Nurbakova
author_sort Johan Erbani
collection DOAJ
description The confusion matrix is a key tool for understanding and evaluating models in supervised classification problems. Various matrices are proposed depending on the problem framework: single-label, multi-label, or even soft-label restricted to probability distributions. However, most of these approaches are not compatible with each other and lack theoretical justification. Leveraging optimal transport theory and the principle of maximum entropy, we propose a unique confusion matrix applicable across single, multi, and soft-label contexts. The Transport-based Confusion Matrix (TCM) extends the classic Confusion Matrix (CM), being identical in the single-label context. TCM introduces a comprehensive, theory-supported description of previously inaccessible errors, thereby enhancing the consistency and scope of machine learning evaluation.
format Article
id doaj-art-6f1e904c9cc6491ca31a2ddd56cc6ca6
institution OA Journals
issn 2169-3536
language English
publishDate 2024-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-6f1e904c9cc6491ca31a2ddd56cc6ca62025-08-20T01:55:52ZengIEEEIEEE Access2169-35362024-01-011218137218141910.1109/ACCESS.2024.350719910769075Confusion Matrices: A Unified TheoryJohan Erbani0https://orcid.org/0009-0000-2717-608XPierre-Edouard Portier1https://orcid.org/0000-0002-6439-9466Elod Egyed-Zsigmond2https://orcid.org/0000-0002-1218-8026Diana Nurbakova3https://orcid.org/0000-0002-6620-7771INSA Lyon, CNRS, UCBL, LIRIS, UMR 5205, Université de Lyon, Villeurbanne, FranceCaisse d’Epargne Rhône-Alpes, Paris, FranceINSA Lyon, CNRS, UCBL, LIRIS, UMR 5205, Université de Lyon, Villeurbanne, FranceINSA Lyon, CNRS, UCBL, LIRIS, UMR 5205, Université de Lyon, Villeurbanne, FranceThe confusion matrix is a key tool for understanding and evaluating models in supervised classification problems. Various matrices are proposed depending on the problem framework: single-label, multi-label, or even soft-label restricted to probability distributions. However, most of these approaches are not compatible with each other and lack theoretical justification. Leveraging optimal transport theory and the principle of maximum entropy, we propose a unique confusion matrix applicable across single, multi, and soft-label contexts. The Transport-based Confusion Matrix (TCM) extends the classic Confusion Matrix (CM), being identical in the single-label context. TCM introduces a comprehensive, theory-supported description of previously inaccessible errors, thereby enhancing the consistency and scope of machine learning evaluation.https://ieeexplore.ieee.org/document/10769075/Classificationevaluationmachine learningmulti-label confusion matrixoptimal transportsingle-label confusion matrix
spellingShingle Johan Erbani
Pierre-Edouard Portier
Elod Egyed-Zsigmond
Diana Nurbakova
Confusion Matrices: A Unified Theory
IEEE Access
Classification
evaluation
machine learning
multi-label confusion matrix
optimal transport
single-label confusion matrix
title Confusion Matrices: A Unified Theory
title_full Confusion Matrices: A Unified Theory
title_fullStr Confusion Matrices: A Unified Theory
title_full_unstemmed Confusion Matrices: A Unified Theory
title_short Confusion Matrices: A Unified Theory
title_sort confusion matrices a unified theory
topic Classification
evaluation
machine learning
multi-label confusion matrix
optimal transport
single-label confusion matrix
url https://ieeexplore.ieee.org/document/10769075/
work_keys_str_mv AT johanerbani confusionmatricesaunifiedtheory
AT pierreedouardportier confusionmatricesaunifiedtheory
AT elodegyedzsigmond confusionmatricesaunifiedtheory
AT diananurbakova confusionmatricesaunifiedtheory