Fidelity loss in distribution-preserving anonymization and histogram equalization

Lav R. Varshney; Kush R. Varshney

doi:10.1109/CISS.2016.7460471

Fidelity loss in distribution-preserving anonymization and histogram equalization

Lav R. Varshney
, Kush R. Varshney

Office of the Provost

IBM

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

Abstract

In this paper, we show a formal equivalence between histogram equalization and distribution-preserving quantization. We use this equivalence to connect histogram equalization to quantization for preserving anonymity under the k-anonymity metric, while maintaining distributional properties for data analytics applications. Finally, we make connections to mismatched quantization. These relationships allow us to characterize the loss in mean-squared error (MSE) performance of privacy-preserving quantizers that must meet distribution-preservation constraints as compared to MSE-optimal quantizers in the high-rate regime. Thus, we obtain a formal characterization of the cost of anonymity.

Original language	English
Title of host publication	2016 50th Annual Conference on Information Systems and Sciences, CISS 2016
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	24-29
Number of pages	6
ISBN (Electronic)	9781467394574
DOIs	https://doi.org/10.1109/CISS.2016.7460471
State	Published - Apr 26 2016
Event	50th Annual Conference on Information Systems and Sciences, CISS 2016 - Princeton, United States Duration: Mar 16 2016 → Mar 18 2016

Publication series

Name	2016 50th Annual Conference on Information Systems and Sciences, CISS 2016

Conference

Conference	50th Annual Conference on Information Systems and Sciences, CISS 2016
Country/Territory	United States
City	Princeton
Period	03/16/16 → 03/18/16

Keywords

Data
Histogram equalization
K-anonymity
Mismatched quantization
Neural information processing

Access to Document

10.1109/CISS.2016.7460471

Cite this

Varshney, L. R., & Varshney, K. R. (2016). Fidelity loss in distribution-preserving anonymization and histogram equalization. In 2016 50th Annual Conference on Information Systems and Sciences, CISS 2016 (pp. 24-29). Article 7460471 (2016 50th Annual Conference on Information Systems and Sciences, CISS 2016). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/CISS.2016.7460471

@inproceedings{2e5150ea1bbc4134b0c1af9ddd2ea7b9,

title = "Fidelity loss in distribution-preserving anonymization and histogram equalization",

abstract = "In this paper, we show a formal equivalence between histogram equalization and distribution-preserving quantization. We use this equivalence to connect histogram equalization to quantization for preserving anonymity under the k-anonymity metric, while maintaining distributional properties for data analytics applications. Finally, we make connections to mismatched quantization. These relationships allow us to characterize the loss in mean-squared error (MSE) performance of privacy-preserving quantizers that must meet distribution-preservation constraints as compared to MSE-optimal quantizers in the high-rate regime. Thus, we obtain a formal characterization of the cost of anonymity.",

keywords = "Data, Histogram equalization, K-anonymity, Mismatched quantization, Neural information processing",

author = "Varshney, \{Lav R.\} and Varshney, \{Kush R.\}",

note = "Publisher Copyright: {\textcopyright} 2016 IEEE.; 50th Annual Conference on Information Systems and Sciences, CISS 2016 ; Conference date: 16-03-2016 Through 18-03-2016",

year = "2016",

month = apr,

day = "26",

doi = "10.1109/CISS.2016.7460471",

language = "English",

series = "2016 50th Annual Conference on Information Systems and Sciences, CISS 2016",

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

pages = "24--29",

booktitle = "2016 50th Annual Conference on Information Systems and Sciences, CISS 2016",

}

Varshney, LR & Varshney, KR 2016, Fidelity loss in distribution-preserving anonymization and histogram equalization. in 2016 50th Annual Conference on Information Systems and Sciences, CISS 2016., 7460471, 2016 50th Annual Conference on Information Systems and Sciences, CISS 2016, Institute of Electrical and Electronics Engineers Inc., pp. 24-29, 50th Annual Conference on Information Systems and Sciences, CISS 2016, Princeton, United States, 03/16/16. https://doi.org/10.1109/CISS.2016.7460471

Fidelity loss in distribution-preserving anonymization and histogram equalization. / Varshney, Lav R.; Varshney, Kush R.
2016 50th Annual Conference on Information Systems and Sciences, CISS 2016. Institute of Electrical and Electronics Engineers Inc., 2016. p. 24-29 7460471 (2016 50th Annual Conference on Information Systems and Sciences, CISS 2016).

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

TY - GEN

T1 - Fidelity loss in distribution-preserving anonymization and histogram equalization

AU - Varshney, Lav R.

AU - Varshney, Kush R.

PY - 2016/4/26

Y1 - 2016/4/26

N2 - In this paper, we show a formal equivalence between histogram equalization and distribution-preserving quantization. We use this equivalence to connect histogram equalization to quantization for preserving anonymity under the k-anonymity metric, while maintaining distributional properties for data analytics applications. Finally, we make connections to mismatched quantization. These relationships allow us to characterize the loss in mean-squared error (MSE) performance of privacy-preserving quantizers that must meet distribution-preservation constraints as compared to MSE-optimal quantizers in the high-rate regime. Thus, we obtain a formal characterization of the cost of anonymity.

AB - In this paper, we show a formal equivalence between histogram equalization and distribution-preserving quantization. We use this equivalence to connect histogram equalization to quantization for preserving anonymity under the k-anonymity metric, while maintaining distributional properties for data analytics applications. Finally, we make connections to mismatched quantization. These relationships allow us to characterize the loss in mean-squared error (MSE) performance of privacy-preserving quantizers that must meet distribution-preservation constraints as compared to MSE-optimal quantizers in the high-rate regime. Thus, we obtain a formal characterization of the cost of anonymity.

KW - Data

KW - Histogram equalization

KW - K-anonymity

KW - Mismatched quantization

KW - Neural information processing

UR - https://www.scopus.com/pages/publications/84992379385

U2 - 10.1109/CISS.2016.7460471

DO - 10.1109/CISS.2016.7460471

M3 - Conference contribution

AN - SCOPUS:84992379385

T3 - 2016 50th Annual Conference on Information Systems and Sciences, CISS 2016

SP - 24

EP - 29

BT - 2016 50th Annual Conference on Information Systems and Sciences, CISS 2016

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 50th Annual Conference on Information Systems and Sciences, CISS 2016

Y2 - 16 March 2016 through 18 March 2016

ER -

Varshney LR, Varshney KR. Fidelity loss in distribution-preserving anonymization and histogram equalization. In 2016 50th Annual Conference on Information Systems and Sciences, CISS 2016. Institute of Electrical and Electronics Engineers Inc. 2016. p. 24-29. 7460471. (2016 50th Annual Conference on Information Systems and Sciences, CISS 2016). doi: 10.1109/CISS.2016.7460471

Fidelity loss in distribution-preserving anonymization and histogram equalization

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this