Improved PAC-Bayesian Bounds for Linear Regression
Document type :
Communication dans un congrès avec actes
Permalink :
Title :
Improved PAC-Bayesian Bounds for Linear Regression
Author(s) :
Shalaeva, Vera [Auteur]
Fakhrizadeh Esfahani, Alireza [Auteur]
Germain, Pascal [Auteur]
Petreczky, Mihaly [Auteur]
Centre de Recherche en Informatique, Signal et Automatique de Lille - UMR 9189 [CRIStAL]
Fakhrizadeh Esfahani, Alireza [Auteur]
Germain, Pascal [Auteur]
Petreczky, Mihaly [Auteur]

Centre de Recherche en Informatique, Signal et Automatique de Lille - UMR 9189 [CRIStAL]
Conference title :
Thirty-Fourth AAAI Conference on Artificial Intelligence
City :
New York
Country :
Etats-Unis d'Amérique
Start date of the conference :
2020-02-07
Publication date :
2020-02-07
HAL domain(s) :
Informatique [cs]/Apprentissage [cs.LG]
English abstract : [en]
In this paper, we improve the PAC-Bayesian error bound for linear regression derived in Germain et al. [10]. The improvements are twofold. First, the proposed error bound is tighter, and converges to the generalization ...
Show more >In this paper, we improve the PAC-Bayesian error bound for linear regression derived in Germain et al. [10]. The improvements are twofold. First, the proposed error bound is tighter, and converges to the generalization loss with a well-chosen temperature parameter. Second, the error bound also holds for training data that are not independently sampled. In particular, the error bound applies to certain time series generated by well-known classes of dynamical models, such as ARX models.Show less >
Show more >In this paper, we improve the PAC-Bayesian error bound for linear regression derived in Germain et al. [10]. The improvements are twofold. First, the proposed error bound is tighter, and converges to the generalization loss with a well-chosen temperature parameter. Second, the error bound also holds for training data that are not independently sampled. In particular, the error bound applies to certain time series generated by well-known classes of dynamical models, such as ARX models.Show less >
Language :
Anglais
Audience :
Internationale
Popular science :
Non
Administrative institution(s) :
CNRS
Université de Lille
Université de Lille
Submission date :
2020-06-08T14:10:44Z
2020-06-09T09:04:13Z
2021-05-20T12:22:45Z
2020-06-09T09:04:13Z
2021-05-20T12:22:45Z
Files
- documen
- Open access
- Access the document