R2FP: Rich and robust feature pooling for mining visual data

Wei Xiong; Bo Du; Lefei Zhang; Ruimin Hu; Wei Bian; Jialie Shen; Dacheng Tao

doi:10.1109/ICDM.2015.98

R2FP: Rich and robust feature pooling for mining visual data

Wei Xiong, Bo Du, Lefei Zhang, Ruimin Hu, Wei Bian, Jialie Shen, Dacheng Tao

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

Abstract

The human visual system proves smart in extracting both global and local features. Can we design a similar way for unsupervised feature learning? In this paper, we propose anovel pooling method within an unsupervised feature learningframework, named Rich and Robust Feature Pooling (R2FP), to better explore rich and robust representation from sparsefeature maps of the input data. Both local and global poolingstrategies are further considered to instantiate such a methodand intensively studied. The former selects the most conductivefeatures in the sub-region and summarizes the joint distributionof the selected features, while the latter is utilized to extractmultiple resolutions of features and fuse the features witha feature balancing kernel for rich representation. Extensiveexperiments on several image recognition tasks demonstratethe superiority of the proposed techniques.

Original language	English
Title of host publication	Proceedings - 15th IEEE International Conference on Data Mining, ICDM 2015
Place of Publication	Atlantic City; United States
Publisher	IEEE, Institute of Electrical and Electronics Engineers
Pages	469-478
Number of pages	10
ISBN (Electronic)	9781467395038
DOIs	https://doi.org/10.1109/ICDM.2015.98
Publication status	Published - Nov 2015
Externally published	Yes
Event	15th IEEE International Conference on Data Mining, ICDM 2015 - Atlantic City, United States Duration: 14 Nov 2015 → 17 Nov 2015

Conference

Conference	15th IEEE International Conference on Data Mining, ICDM 2015
Country/Territory	United States
City	Atlantic City
Period	14/11/15 → 17/11/15

Access to Document

10.1109/ICDM.2015.98

Cite this

@inproceedings{93a35cac684e48fe9f868edeae2a70e5,

title = "R2FP: Rich and robust feature pooling for mining visual data",

abstract = "The human visual system proves smart in extracting both global and local features. Can we design a similar way for unsupervised feature learning? In this paper, we propose anovel pooling method within an unsupervised feature learningframework, named Rich and Robust Feature Pooling (R2FP), to better explore rich and robust representation from sparsefeature maps of the input data. Both local and global poolingstrategies are further considered to instantiate such a methodand intensively studied. The former selects the most conductivefeatures in the sub-region and summarizes the joint distributionof the selected features, while the latter is utilized to extractmultiple resolutions of features and fuse the features witha feature balancing kernel for rich representation. Extensiveexperiments on several image recognition tasks demonstratethe superiority of the proposed techniques.",

keywords = "Autoencoder, Pooling, Representation learning",

author = "Wei Xiong and Bo Du and Lefei Zhang and Ruimin Hu and Wei Bian and Jialie Shen and Dacheng Tao",

year = "2015",

month = nov,

doi = "10.1109/ICDM.2015.98",

language = "English",

pages = "469--478",

booktitle = "Proceedings - 15th IEEE International Conference on Data Mining, ICDM 2015",

publisher = "IEEE, Institute of Electrical and Electronics Engineers",

address = "United States",

note = "15th IEEE International Conference on Data Mining, ICDM 2015 ; Conference date: 14-11-2015 Through 17-11-2015",

}

Xiong, W, Du, B, Zhang, L, Hu, R, Bian, W, Shen, J & Tao, D 2015, R2FP: Rich and robust feature pooling for mining visual data. in Proceedings - 15th IEEE International Conference on Data Mining, ICDM 2015., 15699684, IEEE, Institute of Electrical and Electronics Engineers, Atlantic City; United States, pp. 469-478, 15th IEEE International Conference on Data Mining, ICDM 2015, Atlantic City, United States, 14/11/15. https://doi.org/10.1109/ICDM.2015.98

R2FP : Rich and robust feature pooling for mining visual data. / Xiong, Wei; Du, Bo; Zhang, Lefei et al.

Proceedings - 15th IEEE International Conference on Data Mining, ICDM 2015. Atlantic City; United States : IEEE, Institute of Electrical and Electronics Engineers, 2015. p. 469-478 15699684.

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

TY - GEN

T1 - R2FP

T2 - 15th IEEE International Conference on Data Mining, ICDM 2015

AU - Xiong, Wei

AU - Du, Bo

AU - Zhang, Lefei

AU - Hu, Ruimin

AU - Bian, Wei

AU - Shen, Jialie

AU - Tao, Dacheng

PY - 2015/11

Y1 - 2015/11

N2 - The human visual system proves smart in extracting both global and local features. Can we design a similar way for unsupervised feature learning? In this paper, we propose anovel pooling method within an unsupervised feature learningframework, named Rich and Robust Feature Pooling (R2FP), to better explore rich and robust representation from sparsefeature maps of the input data. Both local and global poolingstrategies are further considered to instantiate such a methodand intensively studied. The former selects the most conductivefeatures in the sub-region and summarizes the joint distributionof the selected features, while the latter is utilized to extractmultiple resolutions of features and fuse the features witha feature balancing kernel for rich representation. Extensiveexperiments on several image recognition tasks demonstratethe superiority of the proposed techniques.

AB - The human visual system proves smart in extracting both global and local features. Can we design a similar way for unsupervised feature learning? In this paper, we propose anovel pooling method within an unsupervised feature learningframework, named Rich and Robust Feature Pooling (R2FP), to better explore rich and robust representation from sparsefeature maps of the input data. Both local and global poolingstrategies are further considered to instantiate such a methodand intensively studied. The former selects the most conductivefeatures in the sub-region and summarizes the joint distributionof the selected features, while the latter is utilized to extractmultiple resolutions of features and fuse the features witha feature balancing kernel for rich representation. Extensiveexperiments on several image recognition tasks demonstratethe superiority of the proposed techniques.

KW - Autoencoder

KW - Pooling

KW - Representation learning

UR - http://www.scopus.com/inward/record.url?scp=84963542743&partnerID=8YFLogxK

U2 - 10.1109/ICDM.2015.98

DO - 10.1109/ICDM.2015.98

M3 - Conference Paper published in Proceedings

SP - 469

EP - 478

BT - Proceedings - 15th IEEE International Conference on Data Mining, ICDM 2015

PB - IEEE, Institute of Electrical and Electronics Engineers

CY - Atlantic City; United States

Y2 - 14 November 2015 through 17 November 2015

ER -

R2FP: Rich and robust feature pooling for mining visual data

Abstract

Conference

Access to Document

Other files and links

Fingerprint

Cite this