Softmaxout transformation-permutation network for facial template protection

Hakyoung Lee; Cheng Yaw Low; Andrew Beng Jin Teoh

doi:10.1109/ICPR48806.2021.9413163

Softmaxout transformation-permutation network for facial template protection

Hakyoung Lee, Cheng Yaw Low, Andrew Beng Jin Teoh

Department of Electrical and Electronic Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

5 Citations (Scopus)

Abstract

In this paper, we propose a data-driven cancellable biometrics scheme, referred to as SoftmaxOut Transformation-Permutation Network (SOTPN). The SOTPN is a neural version of Random Permutation Maxout (RPM) transform, which was introduced for facial template protection. We present a specialized SoftmaxOut layer integrated with the permutable MaxOut units and the parameterized softmax function to approximate the non-differentiable permutation and the winner-takes-all operations in the RPM transform. On top of that, a novel pairwise ArcFace loss and a code balancing loss are also formulated to ensure that the SOTPN-transformed facial template is cancellable, discriminative, high entropy and free from quantization errors when coupled with the SoftmaxOut layer. The proposed SOTPN is evaluated on three face datasets, namely LFW, YouTube Face and Facescrub, and our experimental results disclosed that the SOTPN outperforms the RPM transform significantly.

Original language	English
Title of host publication	Proceedings of ICPR 2020 - 25th International Conference on Pattern Recognition
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	7558-7565
Number of pages	8
ISBN (Electronic)	9781728188089
DOIs	https://doi.org/10.1109/ICPR48806.2021.9413163
Publication status	Published - 2020
Event	25th International Conference on Pattern Recognition, ICPR 2020 - Virtual, Milan, Italy Duration: 2021 Jan 10 → 2021 Jan 15

Publication series

Name	Proceedings - International Conference on Pattern Recognition
ISSN (Print)	1051-4651

Conference

Conference	25th International Conference on Pattern Recognition, ICPR 2020
Country/Territory	Italy
City	Virtual, Milan
Period	21/1/10 → 21/1/15

Bibliographical note

Funding Information:
ACKNOWLEDGEMENT This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) (NO. NRF-2019R1A2C1003306)

Publisher Copyright:
© 2020 IEEE

All Science Journal Classification (ASJC) codes

Computer Vision and Pattern Recognition

Access to Document

10.1109/ICPR48806.2021.9413163

Cite this

Lee, H., Low, C. Y., & Teoh, A. B. J. (2020). Softmaxout transformation-permutation network for facial template protection. In Proceedings of ICPR 2020 - 25th International Conference on Pattern Recognition (pp. 7558-7565). [9413163] (Proceedings - International Conference on Pattern Recognition). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICPR48806.2021.9413163

@inproceedings{82d6fb75692941f4a7baa7b5a26decb8,

title = "Softmaxout transformation-permutation network for facial template protection",

abstract = "In this paper, we propose a data-driven cancellable biometrics scheme, referred to as SoftmaxOut Transformation-Permutation Network (SOTPN). The SOTPN is a neural version of Random Permutation Maxout (RPM) transform, which was introduced for facial template protection. We present a specialized SoftmaxOut layer integrated with the permutable MaxOut units and the parameterized softmax function to approximate the non-differentiable permutation and the winner-takes-all operations in the RPM transform. On top of that, a novel pairwise ArcFace loss and a code balancing loss are also formulated to ensure that the SOTPN-transformed facial template is cancellable, discriminative, high entropy and free from quantization errors when coupled with the SoftmaxOut layer. The proposed SOTPN is evaluated on three face datasets, namely LFW, YouTube Face and Facescrub, and our experimental results disclosed that the SOTPN outperforms the RPM transform significantly.",

author = "Hakyoung Lee and Low, {Cheng Yaw} and Teoh, {Andrew Beng Jin}",

note = "Funding Information: ACKNOWLEDGEMENT This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) (NO. NRF-2019R1A2C1003306) Publisher Copyright: {\textcopyright} 2020 IEEE; 25th International Conference on Pattern Recognition, ICPR 2020 ; Conference date: 10-01-2021 Through 15-01-2021",

year = "2020",

doi = "10.1109/ICPR48806.2021.9413163",

language = "English",

series = "Proceedings - International Conference on Pattern Recognition",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "7558--7565",

booktitle = "Proceedings of ICPR 2020 - 25th International Conference on Pattern Recognition",

address = "United States",

}

Lee, H, Low, CY & Teoh, ABJ 2020, Softmaxout transformation-permutation network for facial template protection. in Proceedings of ICPR 2020 - 25th International Conference on Pattern Recognition., 9413163, Proceedings - International Conference on Pattern Recognition, Institute of Electrical and Electronics Engineers Inc., pp. 7558-7565, 25th International Conference on Pattern Recognition, ICPR 2020, Virtual, Milan, Italy, 21/1/10. https://doi.org/10.1109/ICPR48806.2021.9413163

Softmaxout transformation-permutation network for facial template protection. / Lee, Hakyoung; Low, Cheng Yaw; Teoh, Andrew Beng Jin.

Proceedings of ICPR 2020 - 25th International Conference on Pattern Recognition. Institute of Electrical and Electronics Engineers Inc., 2020. p. 7558-7565 9413163 (Proceedings - International Conference on Pattern Recognition).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - Softmaxout transformation-permutation network for facial template protection

AU - Lee, Hakyoung

AU - Low, Cheng Yaw

AU - Teoh, Andrew Beng Jin

N1 - Funding Information: ACKNOWLEDGEMENT This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) (NO. NRF-2019R1A2C1003306) Publisher Copyright: © 2020 IEEE

PY - 2020

Y1 - 2020

N2 - In this paper, we propose a data-driven cancellable biometrics scheme, referred to as SoftmaxOut Transformation-Permutation Network (SOTPN). The SOTPN is a neural version of Random Permutation Maxout (RPM) transform, which was introduced for facial template protection. We present a specialized SoftmaxOut layer integrated with the permutable MaxOut units and the parameterized softmax function to approximate the non-differentiable permutation and the winner-takes-all operations in the RPM transform. On top of that, a novel pairwise ArcFace loss and a code balancing loss are also formulated to ensure that the SOTPN-transformed facial template is cancellable, discriminative, high entropy and free from quantization errors when coupled with the SoftmaxOut layer. The proposed SOTPN is evaluated on three face datasets, namely LFW, YouTube Face and Facescrub, and our experimental results disclosed that the SOTPN outperforms the RPM transform significantly.

AB - In this paper, we propose a data-driven cancellable biometrics scheme, referred to as SoftmaxOut Transformation-Permutation Network (SOTPN). The SOTPN is a neural version of Random Permutation Maxout (RPM) transform, which was introduced for facial template protection. We present a specialized SoftmaxOut layer integrated with the permutable MaxOut units and the parameterized softmax function to approximate the non-differentiable permutation and the winner-takes-all operations in the RPM transform. On top of that, a novel pairwise ArcFace loss and a code balancing loss are also formulated to ensure that the SOTPN-transformed facial template is cancellable, discriminative, high entropy and free from quantization errors when coupled with the SoftmaxOut layer. The proposed SOTPN is evaluated on three face datasets, namely LFW, YouTube Face and Facescrub, and our experimental results disclosed that the SOTPN outperforms the RPM transform significantly.

UR - http://www.scopus.com/inward/record.url?scp=85110536505&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85110536505&partnerID=8YFLogxK

U2 - 10.1109/ICPR48806.2021.9413163

DO - 10.1109/ICPR48806.2021.9413163

M3 - Conference contribution

AN - SCOPUS:85110536505

T3 - Proceedings - International Conference on Pattern Recognition

SP - 7558

EP - 7565

BT - Proceedings of ICPR 2020 - 25th International Conference on Pattern Recognition

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 25th International Conference on Pattern Recognition, ICPR 2020

Y2 - 10 January 2021 through 15 January 2021

ER -

Lee H, Low CY, Teoh ABJ. Softmaxout transformation-permutation network for facial template protection. In Proceedings of ICPR 2020 - 25th International Conference on Pattern Recognition. Institute of Electrical and Electronics Engineers Inc. 2020. p. 7558-7565. 9413163. (Proceedings - International Conference on Pattern Recognition). doi: 10.1109/ICPR48806.2021.9413163

Softmaxout transformation-permutation network for facial template protection

Abstract

Publication series

Conference

Bibliographical note

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this