A Sharp Lower Bound for Agnostic Learning with Sample Compression Schemes

Steve Hanneke; Aryeh Kontorovich

A Sharp Lower Bound for Agnostic Learning with Sample Compression Schemes

Department of Computer Science

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

Abstract

We establish a tight characterization of the worst-case rates for the excess risk of agnostic learning with sample compression schemes and for uniform convergence for agnostic sample compression schemes. In particular, we find that the optimal rates of convergence for size-$k$ agnostic sample compression schemes are of the form $k n/k)n$, which contrasts with agnostic learning with classes of VC dimension $k$, where the optimal rates are of the form $kn$.

Original language	English GB
Title of host publication	Proceedings of the 30th International Conference on Algorithmic Learning Theory
Editors	Aurélien Garivier, Satyen Kale
Place of Publication	Chicago, Illinois
Publisher	PMLR
Pages	489-505
Number of pages	17
Volume	98
State	Published - 1 Oct 2019

Access to Document

http://proceedings.mlr.press/v98/hanneke19b.html

Cite this

@inproceedings{e5c1c3bfaf4a4e4f8a5eae0e9f60f106,

title = "A Sharp Lower Bound for Agnostic Learning with Sample Compression Schemes",

abstract = "We establish a tight characterization of the worst-case rates for the excess risk of agnostic learning with sample compression schemes and for uniform convergence for agnostic sample compression schemes. In particular, we find that the optimal rates of convergence for size-$k$ agnostic sample compression schemes are of the form $k n/k)n$, which contrasts with agnostic learning with classes of VC dimension $k$, where the optimal rates are of the form $kn$.",

author = "Steve Hanneke and Aryeh Kontorovich",

year = "2019",

month = oct,

day = "1",

language = "אנגלית",

volume = "98",

pages = "489--505",

editor = "Aur{\'e}lien Garivier and Satyen Kale",

booktitle = "Proceedings of the 30th International Conference on Algorithmic Learning Theory",

publisher = "PMLR",

}

A Sharp Lower Bound for Agnostic Learning with Sample Compression Schemes. / Hanneke, Steve; Kontorovich, Aryeh.

Proceedings of the 30th International Conference on Algorithmic Learning Theory. ed. / Aurélien Garivier; Satyen Kale. Vol. 98 Chicago, Illinois : PMLR, 2019. p. 489-505.

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

TY - GEN

T1 - A Sharp Lower Bound for Agnostic Learning with Sample Compression Schemes

AU - Hanneke, Steve

AU - Kontorovich, Aryeh

PY - 2019/10/1

Y1 - 2019/10/1

N2 - We establish a tight characterization of the worst-case rates for the excess risk of agnostic learning with sample compression schemes and for uniform convergence for agnostic sample compression schemes. In particular, we find that the optimal rates of convergence for size-$k$ agnostic sample compression schemes are of the form $k n/k)n$, which contrasts with agnostic learning with classes of VC dimension $k$, where the optimal rates are of the form $kn$.

AB - We establish a tight characterization of the worst-case rates for the excess risk of agnostic learning with sample compression schemes and for uniform convergence for agnostic sample compression schemes. In particular, we find that the optimal rates of convergence for size-$k$ agnostic sample compression schemes are of the form $k n/k)n$, which contrasts with agnostic learning with classes of VC dimension $k$, where the optimal rates are of the form $kn$.

M3 - פרסום בספר כנס

VL - 98

SP - 489

EP - 505

BT - Proceedings of the 30th International Conference on Algorithmic Learning Theory

A2 - Garivier, Aurélien

A2 - Kale, Satyen

PB - PMLR

CY - Chicago, Illinois

ER -

A Sharp Lower Bound for Agnostic Learning with Sample Compression Schemes

Abstract

Access to Document

Fingerprint

Cite this