Sparse Graph Convolutional Networks for Face Recognition

Renjie Wu, Seiichiro Kamata

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

Abstract

In recent years, deep learning networks have substantially improved the performance of face recognition. Although deep learning networks have been very successful, there are limited to underlying Euclidean structure data. When dealing with complex signals such as medical imaging, genetics, social networks and computer vision, recently there has been a growing interest in trying to apply learning on non-Euclidean geometric data. Graph convolutional networks are a new deep learning architecture for analyzing non-Euclidean geometric data. In computer vision, a human face image is modeled as a graph in the irregular domain. A major technical challenge is how to optimize the structured face graph. Because, classification performance critically depends on the quality of the graph. In this paper, we explore an undirected graph convolutional network called SGCNs (k^{3} - sparse graph convolutional networks). The main idea is to use sparsity-constrained optimization that obtain connected sparse subgraphs. A sparse graph of face image is composed of connected sparse subgraphs. Experiments demonstrate that the learned sparse graph has better performance than mutual k-nearest neighbor graph and l1 graph.

Original languageEnglish
Title of host publication2018 15th International Conference on Control, Automation, Robotics and Vision, ICARCV 2018
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages174-179
Number of pages6
ISBN (Electronic)9781538695821
DOIs
Publication statusPublished - 2018 Dec 18
Event15th International Conference on Control, Automation, Robotics and Vision, ICARCV 2018 - Singapore, Singapore
Duration: 2018 Nov 182018 Nov 21

Other

Other15th International Conference on Control, Automation, Robotics and Vision, ICARCV 2018
CountrySingapore
CitySingapore
Period18/11/1818/11/21

Fingerprint

Sparse Graphs
Face recognition
Face Recognition
Computer vision
Graph in graph theory
Face
Computer Vision
Medical imaging
Constrained optimization
Subgraph
Nearest Neighbor Graph
Irregular Domain
Medical Imaging
Constrained Optimization
Sparsity
Undirected Graph
Social Networks
Euclidean
Data Structures
Optimise

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Vision and Pattern Recognition
  • Control and Optimization

Cite this

Wu, R., & Kamata, S. (2018). Sparse Graph Convolutional Networks for Face Recognition. In 2018 15th International Conference on Control, Automation, Robotics and Vision, ICARCV 2018 (pp. 174-179). [8581214] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICARCV.2018.8581214

Sparse Graph Convolutional Networks for Face Recognition. / Wu, Renjie; Kamata, Seiichiro.

2018 15th International Conference on Control, Automation, Robotics and Vision, ICARCV 2018. Institute of Electrical and Electronics Engineers Inc., 2018. p. 174-179 8581214.

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

Wu, R & Kamata, S 2018, Sparse Graph Convolutional Networks for Face Recognition. in 2018 15th International Conference on Control, Automation, Robotics and Vision, ICARCV 2018., 8581214, Institute of Electrical and Electronics Engineers Inc., pp. 174-179, 15th International Conference on Control, Automation, Robotics and Vision, ICARCV 2018, Singapore, Singapore, 18/11/18. https://doi.org/10.1109/ICARCV.2018.8581214
Wu R, Kamata S. Sparse Graph Convolutional Networks for Face Recognition. In 2018 15th International Conference on Control, Automation, Robotics and Vision, ICARCV 2018. Institute of Electrical and Electronics Engineers Inc. 2018. p. 174-179. 8581214 https://doi.org/10.1109/ICARCV.2018.8581214
Wu, Renjie ; Kamata, Seiichiro. / Sparse Graph Convolutional Networks for Face Recognition. 2018 15th International Conference on Control, Automation, Robotics and Vision, ICARCV 2018. Institute of Electrical and Electronics Engineers Inc., 2018. pp. 174-179
@inproceedings{690de5b49f9b4eceb91e61cf67ef3021,
title = "Sparse Graph Convolutional Networks for Face Recognition",
abstract = "In recent years, deep learning networks have substantially improved the performance of face recognition. Although deep learning networks have been very successful, there are limited to underlying Euclidean structure data. When dealing with complex signals such as medical imaging, genetics, social networks and computer vision, recently there has been a growing interest in trying to apply learning on non-Euclidean geometric data. Graph convolutional networks are a new deep learning architecture for analyzing non-Euclidean geometric data. In computer vision, a human face image is modeled as a graph in the irregular domain. A major technical challenge is how to optimize the structured face graph. Because, classification performance critically depends on the quality of the graph. In this paper, we explore an undirected graph convolutional network called SGCNs (k^{3} - sparse graph convolutional networks). The main idea is to use sparsity-constrained optimization that obtain connected sparse subgraphs. A sparse graph of face image is composed of connected sparse subgraphs. Experiments demonstrate that the learned sparse graph has better performance than mutual k-nearest neighbor graph and l1 graph.",
author = "Renjie Wu and Seiichiro Kamata",
year = "2018",
month = "12",
day = "18",
doi = "10.1109/ICARCV.2018.8581214",
language = "English",
pages = "174--179",
booktitle = "2018 15th International Conference on Control, Automation, Robotics and Vision, ICARCV 2018",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - GEN

T1 - Sparse Graph Convolutional Networks for Face Recognition

AU - Wu, Renjie

AU - Kamata, Seiichiro

PY - 2018/12/18

Y1 - 2018/12/18

N2 - In recent years, deep learning networks have substantially improved the performance of face recognition. Although deep learning networks have been very successful, there are limited to underlying Euclidean structure data. When dealing with complex signals such as medical imaging, genetics, social networks and computer vision, recently there has been a growing interest in trying to apply learning on non-Euclidean geometric data. Graph convolutional networks are a new deep learning architecture for analyzing non-Euclidean geometric data. In computer vision, a human face image is modeled as a graph in the irregular domain. A major technical challenge is how to optimize the structured face graph. Because, classification performance critically depends on the quality of the graph. In this paper, we explore an undirected graph convolutional network called SGCNs (k^{3} - sparse graph convolutional networks). The main idea is to use sparsity-constrained optimization that obtain connected sparse subgraphs. A sparse graph of face image is composed of connected sparse subgraphs. Experiments demonstrate that the learned sparse graph has better performance than mutual k-nearest neighbor graph and l1 graph.

AB - In recent years, deep learning networks have substantially improved the performance of face recognition. Although deep learning networks have been very successful, there are limited to underlying Euclidean structure data. When dealing with complex signals such as medical imaging, genetics, social networks and computer vision, recently there has been a growing interest in trying to apply learning on non-Euclidean geometric data. Graph convolutional networks are a new deep learning architecture for analyzing non-Euclidean geometric data. In computer vision, a human face image is modeled as a graph in the irregular domain. A major technical challenge is how to optimize the structured face graph. Because, classification performance critically depends on the quality of the graph. In this paper, we explore an undirected graph convolutional network called SGCNs (k^{3} - sparse graph convolutional networks). The main idea is to use sparsity-constrained optimization that obtain connected sparse subgraphs. A sparse graph of face image is composed of connected sparse subgraphs. Experiments demonstrate that the learned sparse graph has better performance than mutual k-nearest neighbor graph and l1 graph.

UR - http://www.scopus.com/inward/record.url?scp=85060788153&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85060788153&partnerID=8YFLogxK

U2 - 10.1109/ICARCV.2018.8581214

DO - 10.1109/ICARCV.2018.8581214

M3 - Conference contribution

AN - SCOPUS:85060788153

SP - 174

EP - 179

BT - 2018 15th International Conference on Control, Automation, Robotics and Vision, ICARCV 2018

PB - Institute of Electrical and Electronics Engineers Inc.

ER -