Controlling Confusion via Generalisation Bounds
Document type :
Pré-publication ou Document de travail
Title :
Controlling Confusion via Generalisation Bounds
Author(s) :
Adams, Reuben [Auteur]
The Inria London Programme [Inria-London]
Department of Computer science [University College of London] [UCL-CS]
Shawe-Taylor, John [Auteur]
Department of Computer science [University College of London] [UCL-CS]
Guedj, Benjamin [Auteur]
MOdel for Data Analysis and Learning [MODAL]
The Inria London Programme [Inria-London]
The Alan Turing Institute
Department of Computer science [University College of London] [UCL-CS]
The Inria London Programme [Inria-London]
Department of Computer science [University College of London] [UCL-CS]
Shawe-Taylor, John [Auteur]
Department of Computer science [University College of London] [UCL-CS]
Guedj, Benjamin [Auteur]
MOdel for Data Analysis and Learning [MODAL]
The Inria London Programme [Inria-London]
The Alan Turing Institute
Department of Computer science [University College of London] [UCL-CS]
English keyword(s) :
Statistical Learning Theory
PAC-Bayes theory
Classification
Generalisation Bounds
PAC-Bayes theory
Classification
Generalisation Bounds
HAL domain(s) :
Statistiques [stat]/Machine Learning [stat.ML]
Informatique [cs]/Apprentissage [cs.LG]
Statistiques [stat]/Théorie [stat.TH]
Informatique [cs]/Apprentissage [cs.LG]
Statistiques [stat]/Théorie [stat.TH]
English abstract : [en]
We establish new generalisation bounds for multiclass classification by abstracting to a more general setting of discretised error types. Extending the PAC-Bayes theory, we are hence able to provide fine-grained bounds on ...
Show more >We establish new generalisation bounds for multiclass classification by abstracting to a more general setting of discretised error types. Extending the PAC-Bayes theory, we are hence able to provide fine-grained bounds on performance for multiclass classification, as well as applications to other learning problems including discretisation of regression losses. Tractable training objectives are derived from the bounds. The bounds are uniform over all weightings of the discretised error types and thus can be used to bound weightings not foreseen at training, including the full confusion matrix in the multiclass classification case.Show less >
Show more >We establish new generalisation bounds for multiclass classification by abstracting to a more general setting of discretised error types. Extending the PAC-Bayes theory, we are hence able to provide fine-grained bounds on performance for multiclass classification, as well as applications to other learning problems including discretisation of regression losses. Tractable training objectives are derived from the bounds. The bounds are uniform over all weightings of the discretised error types and thus can be used to bound weightings not foreseen at training, including the full confusion matrix in the multiclass classification case.Show less >
Language :
Anglais
Comment :
31 pages
Collections :
Source :
Files
- document
- Open access
- Access the document
- 2202.05560.pdf
- Open access
- Access the document
- 2202.05560
- Open access
- Access the document