Convex-constrained sparse additive modeling and its extensions

Junming Yin, Yaoliang Yu

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Sparse additive modeling is a class of effective methods for performing high-dimensional nonparametric regression. In this work we show how shape constraints such as convexity/ concavity and their extensions, can be integrated into additive models. The proposed sparse difference of convex additive models (SDCAM) can estimate most continuous functions without any a priori smoothness assumption. Motivated by a characterization of difference of convex functions, our method incorporates a natural regularization functional to avoid overfitting and to reduce model complexity. Computationally, we develop an efficient backfitting algorithm with linear periteration complexity. Experiments on both synthetic and real data confirm that our method is competitive against state-of-the-art sparse additive models, with improved performance in most scenarios.

Original languageEnglish (US)
Title of host publicationUncertainty in Artificial Intelligence - Proceedings of the 33rd Conference, UAI 2017
PublisherAUAI Press Corvallis
StatePublished - Jan 1 2017
Event33rd Conference on Uncertainty in Artificial Intelligence, UAI 2017 - Sydney, Australia
Duration: Aug 11 2017Aug 15 2017

Other

Other33rd Conference on Uncertainty in Artificial Intelligence, UAI 2017
CountryAustralia
CitySydney
Period8/11/178/15/17

Fingerprint

Experiments

ASJC Scopus subject areas

  • Artificial Intelligence

Cite this

Yin, J., & Yu, Y. (2017). Convex-constrained sparse additive modeling and its extensions. In Uncertainty in Artificial Intelligence - Proceedings of the 33rd Conference, UAI 2017 AUAI Press Corvallis.

Convex-constrained sparse additive modeling and its extensions. / Yin, Junming; Yu, Yaoliang.

Uncertainty in Artificial Intelligence - Proceedings of the 33rd Conference, UAI 2017. AUAI Press Corvallis, 2017.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Yin, J & Yu, Y 2017, Convex-constrained sparse additive modeling and its extensions. in Uncertainty in Artificial Intelligence - Proceedings of the 33rd Conference, UAI 2017. AUAI Press Corvallis, 33rd Conference on Uncertainty in Artificial Intelligence, UAI 2017, Sydney, Australia, 8/11/17.
Yin J, Yu Y. Convex-constrained sparse additive modeling and its extensions. In Uncertainty in Artificial Intelligence - Proceedings of the 33rd Conference, UAI 2017. AUAI Press Corvallis. 2017
Yin, Junming ; Yu, Yaoliang. / Convex-constrained sparse additive modeling and its extensions. Uncertainty in Artificial Intelligence - Proceedings of the 33rd Conference, UAI 2017. AUAI Press Corvallis, 2017.
@inproceedings{41b52c3185c44f2ca96f194c077bbc5f,
title = "Convex-constrained sparse additive modeling and its extensions",
abstract = "Sparse additive modeling is a class of effective methods for performing high-dimensional nonparametric regression. In this work we show how shape constraints such as convexity/ concavity and their extensions, can be integrated into additive models. The proposed sparse difference of convex additive models (SDCAM) can estimate most continuous functions without any a priori smoothness assumption. Motivated by a characterization of difference of convex functions, our method incorporates a natural regularization functional to avoid overfitting and to reduce model complexity. Computationally, we develop an efficient backfitting algorithm with linear periteration complexity. Experiments on both synthetic and real data confirm that our method is competitive against state-of-the-art sparse additive models, with improved performance in most scenarios.",
author = "Junming Yin and Yaoliang Yu",
year = "2017",
month = "1",
day = "1",
language = "English (US)",
booktitle = "Uncertainty in Artificial Intelligence - Proceedings of the 33rd Conference, UAI 2017",
publisher = "AUAI Press Corvallis",

}

TY - GEN

T1 - Convex-constrained sparse additive modeling and its extensions

AU - Yin, Junming

AU - Yu, Yaoliang

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Sparse additive modeling is a class of effective methods for performing high-dimensional nonparametric regression. In this work we show how shape constraints such as convexity/ concavity and their extensions, can be integrated into additive models. The proposed sparse difference of convex additive models (SDCAM) can estimate most continuous functions without any a priori smoothness assumption. Motivated by a characterization of difference of convex functions, our method incorporates a natural regularization functional to avoid overfitting and to reduce model complexity. Computationally, we develop an efficient backfitting algorithm with linear periteration complexity. Experiments on both synthetic and real data confirm that our method is competitive against state-of-the-art sparse additive models, with improved performance in most scenarios.

AB - Sparse additive modeling is a class of effective methods for performing high-dimensional nonparametric regression. In this work we show how shape constraints such as convexity/ concavity and their extensions, can be integrated into additive models. The proposed sparse difference of convex additive models (SDCAM) can estimate most continuous functions without any a priori smoothness assumption. Motivated by a characterization of difference of convex functions, our method incorporates a natural regularization functional to avoid overfitting and to reduce model complexity. Computationally, we develop an efficient backfitting algorithm with linear periteration complexity. Experiments on both synthetic and real data confirm that our method is competitive against state-of-the-art sparse additive models, with improved performance in most scenarios.

UR - http://www.scopus.com/inward/record.url?scp=85031121923&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85031121923&partnerID=8YFLogxK

M3 - Conference contribution

BT - Uncertainty in Artificial Intelligence - Proceedings of the 33rd Conference, UAI 2017

PB - AUAI Press Corvallis

ER -