On Information-Theoretic Determination of Misspecified Rates of Convergence

Nir Weinberger; Meir Feder

doi:https://doi.org/10.1109/ISIT50566.2022.9834743

On Information-Theoretic Determination of Misspecified Rates of Convergence

Nir Weinberger, Meir Feder

Technion - Israel Institute of Technology, Tel Aviv University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

We consider the problem of learning a model from given data samples in which the predictor's quality is measured by the log loss. We focus on the misspecified setting, in which the true model generating the data is chosen from a set different from the possible models that can be chosen by the learner. We establish minimax expected regret upper and lower bounds in terms of properly defined projected covering and packing entropies, and show their relation to M-projection geometric properties. We exemplify the bounds in a few settings.

Original language	English
Title of host publication	2022 IEEE International Symposium on Information Theory, ISIT 2022
Pages	1695-1700
Number of pages	6
ISBN (Electronic)	9781665421591
DOIs	https://doi.org/10.1109/ISIT50566.2022.9834743
State	Published - 2022
Event	2022 IEEE International Symposium on Information Theory, ISIT 2022 - Espoo, Finland Duration: 26 Jun 2022 → 1 Jul 2022

Publication series

Name	IEEE International Symposium on Information Theory - Proceedings
Volume	2022-June

Conference

Conference	2022 IEEE International Symposium on Information Theory, ISIT 2022
Country/Territory	Finland
City	Espoo
Period	26/06/22 → 1/07/22

All Science Journal Classification (ASJC) codes

Theoretical Computer Science
Information Systems
Modelling and Simulation
Applied Mathematics

Access to Document

https://doi.org/10.1109/ISIT50566.2022.9834743

Cite this

@inproceedings{d3d7b90626ea45608e95a5a6f7b4740d,

title = "On Information-Theoretic Determination of Misspecified Rates of Convergence",

abstract = "We consider the problem of learning a model from given data samples in which the predictor's quality is measured by the log loss. We focus on the misspecified setting, in which the true model generating the data is chosen from a set different from the possible models that can be chosen by the learner. We establish minimax expected regret upper and lower bounds in terms of properly defined projected covering and packing entropies, and show their relation to M-projection geometric properties. We exemplify the bounds in a few settings.",

author = "Nir Weinberger and Meir Feder",

note = "Publisher Copyright: {\textcopyright} 2022 IEEE.; 2022 IEEE International Symposium on Information Theory, ISIT 2022 ; Conference date: 26-06-2022 Through 01-07-2022",

year = "2022",

doi = "https://doi.org/10.1109/ISIT50566.2022.9834743",

language = "الإنجليزيّة",

series = "IEEE International Symposium on Information Theory - Proceedings",

pages = "1695--1700",

booktitle = "2022 IEEE International Symposium on Information Theory, ISIT 2022",

}

Weinberger, N & Feder, M 2022, On Information-Theoretic Determination of Misspecified Rates of Convergence. in 2022 IEEE International Symposium on Information Theory, ISIT 2022. IEEE International Symposium on Information Theory - Proceedings, vol. 2022-June, pp. 1695-1700, 2022 IEEE International Symposium on Information Theory, ISIT 2022, Espoo, Finland, 26/06/22. https://doi.org/10.1109/ISIT50566.2022.9834743

On Information-Theoretic Determination of Misspecified Rates of Convergence. / Weinberger, Nir; Feder, Meir.
2022 IEEE International Symposium on Information Theory, ISIT 2022. 2022. p. 1695-1700 (IEEE International Symposium on Information Theory - Proceedings; Vol. 2022-June).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - On Information-Theoretic Determination of Misspecified Rates of Convergence

AU - Weinberger, Nir

AU - Feder, Meir

PY - 2022

Y1 - 2022

N2 - We consider the problem of learning a model from given data samples in which the predictor's quality is measured by the log loss. We focus on the misspecified setting, in which the true model generating the data is chosen from a set different from the possible models that can be chosen by the learner. We establish minimax expected regret upper and lower bounds in terms of properly defined projected covering and packing entropies, and show their relation to M-projection geometric properties. We exemplify the bounds in a few settings.

AB - We consider the problem of learning a model from given data samples in which the predictor's quality is measured by the log loss. We focus on the misspecified setting, in which the true model generating the data is chosen from a set different from the possible models that can be chosen by the learner. We establish minimax expected regret upper and lower bounds in terms of properly defined projected covering and packing entropies, and show their relation to M-projection geometric properties. We exemplify the bounds in a few settings.

UR - http://www.scopus.com/inward/record.url?scp=85136319015&partnerID=8YFLogxK

U2 - https://doi.org/10.1109/ISIT50566.2022.9834743

DO - https://doi.org/10.1109/ISIT50566.2022.9834743

M3 - منشور من مؤتمر

T3 - IEEE International Symposium on Information Theory - Proceedings

SP - 1695

EP - 1700

BT - 2022 IEEE International Symposium on Information Theory, ISIT 2022

T2 - 2022 IEEE International Symposium on Information Theory, ISIT 2022

Y2 - 26 June 2022 through 1 July 2022

ER -

On Information-Theoretic Determination of Misspecified Rates of Convergence

Abstract

Publication series

Conference

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this