Robust High-Dimensional Linear Discriminant Analysis under Training Data Contamination

Yuyang Shi; Aditya Deshmukh; Yajun Mei; Venugopal Veeravalli

doi:10.1109/ISIT54713.2023.10206749

Robust High-Dimensional Linear Discriminant Analysis under Training Data Contamination

Yuyang Shi
, Aditya Deshmukh
, Yajun Mei
, Venugopal Veeravalli

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

Abstract

The problem of robust Sparse Linear Discriminant Analysis (LDA) in high-dimensions is studied, in which a fraction of the training data may be corrupted by an adversary. A computationally efficient algorithm is proposed by adapting robust mean estimation along with a calibration framework for LDA. Theoretical properties of the proposed algorithm are established for both the estimation error of the optimal projection vector and the mis-classification rate. Results from extensive numerical studies on both synthetic and real datasets are reported to show the usefulness of our algorithm.

Original language	English (US)
Title of host publication	2023 IEEE International Symposium on Information Theory, ISIT 2023
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	2099-2104
Number of pages	6
ISBN (Electronic)	9781665475549
DOIs	https://doi.org/10.1109/ISIT54713.2023.10206749
State	Published - 2023
Event	2023 IEEE International Symposium on Information Theory, ISIT 2023 - Taipei, Taiwan, Province of China Duration: Jun 25 2023 → Jun 30 2023

Publication series

Name	IEEE International Symposium on Information Theory - Proceedings
Volume	2023-June
ISSN (Print)	2157-8095

Conference

Conference	2023 IEEE International Symposium on Information Theory, ISIT 2023
Country/Territory	Taiwan, Province of China
City	Taipei
Period	6/25/23 → 6/30/23

ASJC Scopus subject areas

Theoretical Computer Science
Information Systems
Modeling and Simulation
Applied Mathematics

Online availability

10.1109/ISIT54713.2023.10206749

Library availability

Discover UIUC Full Text

Cite this

Shi, Y., Deshmukh, A., Mei, Y., & Veeravalli, V. (2023). Robust High-Dimensional Linear Discriminant Analysis under Training Data Contamination. In 2023 IEEE International Symposium on Information Theory, ISIT 2023 (pp. 2099-2104). (IEEE International Symposium on Information Theory - Proceedings; Vol. 2023-June). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ISIT54713.2023.10206749

Robust High-Dimensional Linear Discriminant Analysis under Training Data Contamination. / Shi, Yuyang; Deshmukh, Aditya; Mei, Yajun et al.
2023 IEEE International Symposium on Information Theory, ISIT 2023. Institute of Electrical and Electronics Engineers Inc., 2023. p. 2099-2104 (IEEE International Symposium on Information Theory - Proceedings; Vol. 2023-June).

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

Shi, Y, Deshmukh, A, Mei, Y & Veeravalli, V 2023, Robust High-Dimensional Linear Discriminant Analysis under Training Data Contamination. in 2023 IEEE International Symposium on Information Theory, ISIT 2023. IEEE International Symposium on Information Theory - Proceedings, vol. 2023-June, Institute of Electrical and Electronics Engineers Inc., pp. 2099-2104, 2023 IEEE International Symposium on Information Theory, ISIT 2023, Taipei, Taiwan, Province of China, 6/25/23. https://doi.org/10.1109/ISIT54713.2023.10206749

Shi Y, Deshmukh A, Mei Y, Veeravalli V. Robust High-Dimensional Linear Discriminant Analysis under Training Data Contamination. In 2023 IEEE International Symposium on Information Theory, ISIT 2023. Institute of Electrical and Electronics Engineers Inc. 2023. p. 2099-2104. (IEEE International Symposium on Information Theory - Proceedings). doi: 10.1109/ISIT54713.2023.10206749

@inproceedings{10ffcc747dd44b4b87d2d019c9571982,

title = "Robust High-Dimensional Linear Discriminant Analysis under Training Data Contamination",

abstract = "The problem of robust Sparse Linear Discriminant Analysis (LDA) in high-dimensions is studied, in which a fraction of the training data may be corrupted by an adversary. A computationally efficient algorithm is proposed by adapting robust mean estimation along with a calibration framework for LDA. Theoretical properties of the proposed algorithm are established for both the estimation error of the optimal projection vector and the mis-classification rate. Results from extensive numerical studies on both synthetic and real datasets are reported to show the usefulness of our algorithm.",

author = "Yuyang Shi and Aditya Deshmukh and Yajun Mei and Venugopal Veeravalli",

note = "ACKNOWLEDGEMENT Y. Shi and Y. Mei were supported in part by an NSF-DMS grant 2015405 and by an NIH grant 1R21AI157618-01A1. A. Deshmukh and V. Veeravalli were supported in part by an NSF grant 2106727 and the US Army Research Laboratory under Cooperative Agreement W911NF-17-2-0196.; 2023 IEEE International Symposium on Information Theory, ISIT 2023 ; Conference date: 25-06-2023 Through 30-06-2023",

year = "2023",

doi = "10.1109/ISIT54713.2023.10206749",

language = "English (US)",

series = "IEEE International Symposium on Information Theory - Proceedings",

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

pages = "2099--2104",

booktitle = "2023 IEEE International Symposium on Information Theory, ISIT 2023",

address = "United States",

}

TY - GEN

T1 - Robust High-Dimensional Linear Discriminant Analysis under Training Data Contamination

AU - Shi, Yuyang

AU - Deshmukh, Aditya

AU - Mei, Yajun

AU - Veeravalli, Venugopal

N1 - ACKNOWLEDGEMENT Y. Shi and Y. Mei were supported in part by an NSF-DMS grant 2015405 and by an NIH grant 1R21AI157618-01A1. A. Deshmukh and V. Veeravalli were supported in part by an NSF grant 2106727 and the US Army Research Laboratory under Cooperative Agreement W911NF-17-2-0196.

PY - 2023

Y1 - 2023

N2 - The problem of robust Sparse Linear Discriminant Analysis (LDA) in high-dimensions is studied, in which a fraction of the training data may be corrupted by an adversary. A computationally efficient algorithm is proposed by adapting robust mean estimation along with a calibration framework for LDA. Theoretical properties of the proposed algorithm are established for both the estimation error of the optimal projection vector and the mis-classification rate. Results from extensive numerical studies on both synthetic and real datasets are reported to show the usefulness of our algorithm.

AB - The problem of robust Sparse Linear Discriminant Analysis (LDA) in high-dimensions is studied, in which a fraction of the training data may be corrupted by an adversary. A computationally efficient algorithm is proposed by adapting robust mean estimation along with a calibration framework for LDA. Theoretical properties of the proposed algorithm are established for both the estimation error of the optimal projection vector and the mis-classification rate. Results from extensive numerical studies on both synthetic and real datasets are reported to show the usefulness of our algorithm.

UR - https://www.scopus.com/pages/publications/85171447943

UR - https://www.scopus.com/pages/publications/85171447943#tab=citedBy

U2 - 10.1109/ISIT54713.2023.10206749

DO - 10.1109/ISIT54713.2023.10206749

M3 - Conference contribution

AN - SCOPUS:85171447943

T3 - IEEE International Symposium on Information Theory - Proceedings

SP - 2099

EP - 2104

BT - 2023 IEEE International Symposium on Information Theory, ISIT 2023

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2023 IEEE International Symposium on Information Theory, ISIT 2023

Y2 - 25 June 2023 through 30 June 2023

ER -

Robust High-Dimensional Linear Discriminant Analysis under Training Data Contamination

Abstract

Publication series

Conference

ASJC Scopus subject areas

Online availability

Library availability

Related links

Fingerprint

Cite this