Scalable nonlinear learning with adaptive polynomial expansions

Alekh Agarwal; Alina Beygelzimer; Daniel Hsu; John Langford; Matus Telgarsky

Scalable nonlinear learning with adaptive polynomial expansions

Alekh Agarwal, Alina Beygelzimer, Daniel Hsu, John Langford, Matus Telgarsky

Computer Science

Research output: Contribution to journal › Conference article › peer-review

Abstract

Can we effectively learn a nonlinear representation in time comparable to linear learning? We describe a new algorithm that explicitly and adaptively expands higher-order interaction features over base linear representations. The algorithm is designed for extreme computational efficiency, and an extensive experimental study shows that its computation/prediction tradeoff ability compares very favorably against strong baselines.

Original language	English (US)
Pages (from-to)	2051-2059
Number of pages	9
Journal	Advances in Neural Information Processing Systems
Volume	3
Issue number	January
State	Published - 2014
Event	28th Annual Conference on Neural Information Processing Systems 2014, NIPS 2014 - Montreal, Canada Duration: Dec 8 2014 → Dec 13 2014

ASJC Scopus subject areas

Computer Networks and Communications
Information Systems
Signal Processing

Cite this

@article{ab07143e6ebf4c0cbc2eb9cb01e6cc6b,

title = "Scalable nonlinear learning with adaptive polynomial expansions",

abstract = "Can we effectively learn a nonlinear representation in time comparable to linear learning? We describe a new algorithm that explicitly and adaptively expands higher-order interaction features over base linear representations. The algorithm is designed for extreme computational efficiency, and an extensive experimental study shows that its computation/prediction tradeoff ability compares very favorably against strong baselines.",

author = "Alekh Agarwal and Alina Beygelzimer and Daniel Hsu and John Langford and Matus Telgarsky",

year = "2014",

language = "English (US)",

volume = "3",

pages = "2051--2059",

journal = "Advances in Neural Information Processing Systems",

issn = "1049-5258",

number = "January",

note = "28th Annual Conference on Neural Information Processing Systems 2014, NIPS 2014 ; Conference date: 08-12-2014 Through 13-12-2014",

}

TY - JOUR

T1 - Scalable nonlinear learning with adaptive polynomial expansions

AU - Agarwal, Alekh

AU - Beygelzimer, Alina

AU - Hsu, Daniel

AU - Langford, John

AU - Telgarsky, Matus

PY - 2014

Y1 - 2014

N2 - Can we effectively learn a nonlinear representation in time comparable to linear learning? We describe a new algorithm that explicitly and adaptively expands higher-order interaction features over base linear representations. The algorithm is designed for extreme computational efficiency, and an extensive experimental study shows that its computation/prediction tradeoff ability compares very favorably against strong baselines.

AB - Can we effectively learn a nonlinear representation in time comparable to linear learning? We describe a new algorithm that explicitly and adaptively expands higher-order interaction features over base linear representations. The algorithm is designed for extreme computational efficiency, and an extensive experimental study shows that its computation/prediction tradeoff ability compares very favorably against strong baselines.

UR - http://www.scopus.com/inward/record.url?scp=84937935486&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84937935486&partnerID=8YFLogxK

M3 - Conference article

AN - SCOPUS:84937935486

SN - 1049-5258

VL - 3

SP - 2051

EP - 2059

JO - Advances in Neural Information Processing Systems

JF - Advances in Neural Information Processing Systems

IS - January

T2 - 28th Annual Conference on Neural Information Processing Systems 2014, NIPS 2014

Y2 - 8 December 2014 through 13 December 2014

ER -

Scalable nonlinear learning with adaptive polynomial expansions

Abstract

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this