Composite Multiclass Losses

Elodie Vernet; Robert Williamson; Mark Reid

Composite Multiclass Losses

Elodie Vernet, Robert Williamson, Mark Reid

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

Abstract

We consider loss functions for multiclass prediction problems. We show when a multiclass loss can be expressed as a "proper composite loss", which is the composition of a proper loss and a link function. We extend existing results for binary losses to mul

Original language	English
Title of host publication	Advances in Neural Information Processing Systems 23
Editors	Rich Zemel, John Shawe-Taylor, Peter Bartlett, Fernando Pereira and Kilian Weinb
Place of Publication	Granada Spain
Publisher	Neural Information Processing Systems Foundation
Pages	15
Edition	Peer Reviewed
ISBN (Print)	9781618395993
Publication status	Published - 2011
Event	Neural Information Processing Systems (NIPS 2011) - Granada Spain Duration: 1 Jan 2011 → … https://papers.nips.cc/paper/4487-contextual-gaussian-process-bandit-optimization

Conference

Conference	Neural Information Processing Systems (NIPS 2011)
Period	1/01/11 → …
Other	December 13-15 2011
Internet address	https://papers.nips.cc/paper/4487-contextual-gaussian-process-bandit-optimization

Cite this

@inproceedings{e273174534df47ce8705c940953af189,

title = "Composite Multiclass Losses",

abstract = "We consider loss functions for multiclass prediction problems. We show when a multiclass loss can be expressed as a {"}proper composite loss{"}, which is the composition of a proper loss and a link function. We extend existing results for binary losses to mul",

author = "Elodie Vernet and Robert Williamson and Mark Reid",

year = "2011",

language = "English",

isbn = "9781618395993",

pages = "15",

editor = "\{Rich Zemel, John Shawe-Taylor, Peter Bartlett, Fernando Pereira and Kilian Weinb\}",

booktitle = "Advances in Neural Information Processing Systems 23",

publisher = "Neural Information Processing Systems Foundation",

edition = "Peer Reviewed",

note = "Neural Information Processing Systems (NIPS 2011) ; Conference date: 01-01-2011",

url = "https://papers.nips.cc/paper/4487-contextual-gaussian-process-bandit-optimization",

}

Composite Multiclass Losses. / Vernet, Elodie; Williamson, Robert; Reid, Mark.
Advances in Neural Information Processing Systems 23. ed. / Rich Zemel, John Shawe-Taylor, Peter Bartlett, Fernando Pereira and Kilian Weinb. Peer Reviewed. ed. Granada Spain: Neural Information Processing Systems Foundation, 2011. p. 15.

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

TY - GEN

T1 - Composite Multiclass Losses

AU - Vernet, Elodie

AU - Williamson, Robert

AU - Reid, Mark

PY - 2011

Y1 - 2011

N2 - We consider loss functions for multiclass prediction problems. We show when a multiclass loss can be expressed as a "proper composite loss", which is the composition of a proper loss and a link function. We extend existing results for binary losses to mul

AB - We consider loss functions for multiclass prediction problems. We show when a multiclass loss can be expressed as a "proper composite loss", which is the composition of a proper loss and a link function. We extend existing results for binary losses to mul

UR - https://www.scopus.com/pages/publications/84889719934

M3 - Conference Paper

SN - 9781618395993

SP - 15

BT - Advances in Neural Information Processing Systems 23

A2 - null, Rich Zemel, John Shawe-Taylor, Peter Bartlett, Fernando Pereira and Kilian Weinb

PB - Neural Information Processing Systems Foundation

CY - Granada Spain

T2 - Neural Information Processing Systems (NIPS 2011)

Y2 - 1 January 2011

ER -

Composite Multiclass Losses

Abstract

Conference

Other files and links

Fingerprint

Cite this