Secure Metric Learning via Differential Pairwise Privacy
© 2005-2012 IEEE. Distance Metric Learning (DML) has drawn much attention over the last two decades. A number of previous works have shown that it performs well in measuring the similarities of individuals given a set of correctly labeled pairwise data by domain experts. These important and precisel...
Main Authors: | , , , |
---|---|
Format: | Article in Journal/Newspaper |
Language: | English |
Published: |
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
2021
|
Subjects: | |
Online Access: | http://hdl.handle.net/10453/147240 |
id |
ftunivtsydney:oai:opus.lib.uts.edu.au:10453/147240 |
---|---|
record_format |
openpolar |
spelling |
ftunivtsydney:oai:opus.lib.uts.edu.au:10453/147240 2023-05-15T16:01:33+02:00 Secure Metric Learning via Differential Pairwise Privacy Li, J Pan, Y Sui, Y Tsang, IW 2021-03-16T03:19:07Z application/pdf http://hdl.handle.net/10453/147240 English eng IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC http://purl.org/au-research/grants/arc/DP180100106 http://purl.org/au-research/grants/arc/DE170101081 http://purl.org/au-research/grants/arc/DP200101328 IEEE Transactions on Information Forensics and Security 10.1109/TIFS.2020.2997183 IEEE Transactions on Information Forensics and Security, 2020, 15, pp. 3640-3652 1556-6013 1556-6021 http://hdl.handle.net/10453/147240 © 2020 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. info:eu-repo/semantics/closedAccess 08 Information and Computing Sciences 09 Engineering Strategic Defence & Security Studies Journal Article 2021 ftunivtsydney 2022-03-13T13:44:08Z © 2005-2012 IEEE. Distance Metric Learning (DML) has drawn much attention over the last two decades. A number of previous works have shown that it performs well in measuring the similarities of individuals given a set of correctly labeled pairwise data by domain experts. These important and precisely-labeled pairwise data are often highly sensitive in real world (e.g., patients similarity). This paper studies, for the first time, how pairwise information can be leaked to attackers during distance metric learning, and develops differential pairwise privacy (DPP), generalizing the definition of standard differential privacy, for secure metric learning. Unlike traditional differential privacy which only applies to independent samples, thus cannot be used for pairwise data, DPP successfully deals with this problem by reformulating the worst case. Specifically, given the pairwise data, we reveal all the involved correlations among pairs in the constructed undirected graph. DPP is then formalized that defines what kind of DML algorithm is private to preserve pairwise data. After that, a case study employing the contrastive loss is exhibited to clarify the details of implementing a DPP- DML algorithm. Particularly, the sensitivity reduction technique is proposed to enhance the utility of the output distance metric. Experiments both on a toy dataset and benchmarks demonstrate that the proposed scheme achieves pairwise data privacy without compromising the output performance much (Accuracy declines less than 0.01 throughout all benchmark datasets when the privacy budget is set at 4). Article in Journal/Newspaper DML University of Technology Sydney: OPUS - Open Publications of UTS Scholars |
institution |
Open Polar |
collection |
University of Technology Sydney: OPUS - Open Publications of UTS Scholars |
op_collection_id |
ftunivtsydney |
language |
English |
topic |
08 Information and Computing Sciences 09 Engineering Strategic Defence & Security Studies |
spellingShingle |
08 Information and Computing Sciences 09 Engineering Strategic Defence & Security Studies Li, J Pan, Y Sui, Y Tsang, IW Secure Metric Learning via Differential Pairwise Privacy |
topic_facet |
08 Information and Computing Sciences 09 Engineering Strategic Defence & Security Studies |
description |
© 2005-2012 IEEE. Distance Metric Learning (DML) has drawn much attention over the last two decades. A number of previous works have shown that it performs well in measuring the similarities of individuals given a set of correctly labeled pairwise data by domain experts. These important and precisely-labeled pairwise data are often highly sensitive in real world (e.g., patients similarity). This paper studies, for the first time, how pairwise information can be leaked to attackers during distance metric learning, and develops differential pairwise privacy (DPP), generalizing the definition of standard differential privacy, for secure metric learning. Unlike traditional differential privacy which only applies to independent samples, thus cannot be used for pairwise data, DPP successfully deals with this problem by reformulating the worst case. Specifically, given the pairwise data, we reveal all the involved correlations among pairs in the constructed undirected graph. DPP is then formalized that defines what kind of DML algorithm is private to preserve pairwise data. After that, a case study employing the contrastive loss is exhibited to clarify the details of implementing a DPP- DML algorithm. Particularly, the sensitivity reduction technique is proposed to enhance the utility of the output distance metric. Experiments both on a toy dataset and benchmarks demonstrate that the proposed scheme achieves pairwise data privacy without compromising the output performance much (Accuracy declines less than 0.01 throughout all benchmark datasets when the privacy budget is set at 4). |
format |
Article in Journal/Newspaper |
author |
Li, J Pan, Y Sui, Y Tsang, IW |
author_facet |
Li, J Pan, Y Sui, Y Tsang, IW |
author_sort |
Li, J |
title |
Secure Metric Learning via Differential Pairwise Privacy |
title_short |
Secure Metric Learning via Differential Pairwise Privacy |
title_full |
Secure Metric Learning via Differential Pairwise Privacy |
title_fullStr |
Secure Metric Learning via Differential Pairwise Privacy |
title_full_unstemmed |
Secure Metric Learning via Differential Pairwise Privacy |
title_sort |
secure metric learning via differential pairwise privacy |
publisher |
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC |
publishDate |
2021 |
url |
http://hdl.handle.net/10453/147240 |
genre |
DML |
genre_facet |
DML |
op_relation |
http://purl.org/au-research/grants/arc/DP180100106 http://purl.org/au-research/grants/arc/DE170101081 http://purl.org/au-research/grants/arc/DP200101328 IEEE Transactions on Information Forensics and Security 10.1109/TIFS.2020.2997183 IEEE Transactions on Information Forensics and Security, 2020, 15, pp. 3640-3652 1556-6013 1556-6021 http://hdl.handle.net/10453/147240 |
op_rights |
© 2020 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. info:eu-repo/semantics/closedAccess |
_version_ |
1766397361784881152 |