Representative Selection with Structured Sparsity

Hongxing Wang, Yoshinobu Kawahara, Chaoqun Weng, Junsong Yuan

Research output: Contribution to journalArticle

16 Citations (Scopus)

Abstract

We propose a novel formulation to find representatives in data samples via learning with structured sparsity. To find representatives with both diversity and representativeness, we formulate the problem as a structurally-regularized learning where the objective function consists of a reconstruction error and three structured regularizers: (1) group sparsity regularizer, (2) diversity regularizer, and (3) locality-sensitivity regularizer. For the optimization of the objective, we propose an accelerated proximal gradient algorithm, combined with the proximal-Dykstra method and the calculation of parametric maximum flows. Experiments on image and video data validate the effectiveness of our method in finding exemplars with diversity and representativeness and demonstrate its robustness to outliers.

Original languageEnglish
Pages (from-to)268-278
Number of pages11
JournalPattern Recognition
Volume63
DOIs
Publication statusPublished - Mar 1 2017

Fingerprint

Experiments

All Science Journal Classification (ASJC) codes

  • Software
  • Signal Processing
  • Computer Vision and Pattern Recognition
  • Artificial Intelligence

Cite this

Representative Selection with Structured Sparsity. / Wang, Hongxing; Kawahara, Yoshinobu; Weng, Chaoqun; Yuan, Junsong.

In: Pattern Recognition, Vol. 63, 01.03.2017, p. 268-278.

Research output: Contribution to journalArticle

Wang, Hongxing ; Kawahara, Yoshinobu ; Weng, Chaoqun ; Yuan, Junsong. / Representative Selection with Structured Sparsity. In: Pattern Recognition. 2017 ; Vol. 63. pp. 268-278.
@article{600eb414fd2644988ba5fbd0fabefc78,
title = "Representative Selection with Structured Sparsity",
abstract = "We propose a novel formulation to find representatives in data samples via learning with structured sparsity. To find representatives with both diversity and representativeness, we formulate the problem as a structurally-regularized learning where the objective function consists of a reconstruction error and three structured regularizers: (1) group sparsity regularizer, (2) diversity regularizer, and (3) locality-sensitivity regularizer. For the optimization of the objective, we propose an accelerated proximal gradient algorithm, combined with the proximal-Dykstra method and the calculation of parametric maximum flows. Experiments on image and video data validate the effectiveness of our method in finding exemplars with diversity and representativeness and demonstrate its robustness to outliers.",
author = "Hongxing Wang and Yoshinobu Kawahara and Chaoqun Weng and Junsong Yuan",
year = "2017",
month = "3",
day = "1",
doi = "10.1016/j.patcog.2016.10.014",
language = "English",
volume = "63",
pages = "268--278",
journal = "Pattern Recognition",
issn = "0031-3203",
publisher = "Elsevier Limited",

}

TY - JOUR

T1 - Representative Selection with Structured Sparsity

AU - Wang, Hongxing

AU - Kawahara, Yoshinobu

AU - Weng, Chaoqun

AU - Yuan, Junsong

PY - 2017/3/1

Y1 - 2017/3/1

N2 - We propose a novel formulation to find representatives in data samples via learning with structured sparsity. To find representatives with both diversity and representativeness, we formulate the problem as a structurally-regularized learning where the objective function consists of a reconstruction error and three structured regularizers: (1) group sparsity regularizer, (2) diversity regularizer, and (3) locality-sensitivity regularizer. For the optimization of the objective, we propose an accelerated proximal gradient algorithm, combined with the proximal-Dykstra method and the calculation of parametric maximum flows. Experiments on image and video data validate the effectiveness of our method in finding exemplars with diversity and representativeness and demonstrate its robustness to outliers.

AB - We propose a novel formulation to find representatives in data samples via learning with structured sparsity. To find representatives with both diversity and representativeness, we formulate the problem as a structurally-regularized learning where the objective function consists of a reconstruction error and three structured regularizers: (1) group sparsity regularizer, (2) diversity regularizer, and (3) locality-sensitivity regularizer. For the optimization of the objective, we propose an accelerated proximal gradient algorithm, combined with the proximal-Dykstra method and the calculation of parametric maximum flows. Experiments on image and video data validate the effectiveness of our method in finding exemplars with diversity and representativeness and demonstrate its robustness to outliers.

UR - http://www.scopus.com/inward/record.url?scp=84998953433&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84998953433&partnerID=8YFLogxK

U2 - 10.1016/j.patcog.2016.10.014

DO - 10.1016/j.patcog.2016.10.014

M3 - Article

VL - 63

SP - 268

EP - 278

JO - Pattern Recognition

JF - Pattern Recognition

SN - 0031-3203

ER -