Adversarial Training for Probabilistic Spiking Neural Networks

Alireza Bagheri; Osvaldo Simeone; Bipin Rajendran

doi:10.1109/SPAWC.2018.8446003

Adversarial Training for Probabilistic Spiking Neural Networks

Alireza Bagheri, Osvaldo Simeone, Bipin Rajendran

Electrical and Computer Engineering

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

23 Scopus citations

Abstract

Classifiers trained using conventional empirical risk minimization or maximum likelihood methods are known to suffer dramatic performance degradations when tested over examples adversarially selected based on knowledge of the clas-sifier's decision rule. Due to the prominence of Artificial Neural Networks (ANNs) as classifiers, their sensitivity to adversarial examples, as well as robust training schemes, have been recently the subject of intense investigation. In this paper, for the first time, the sensitivity of spiking neural networks (SNNs), or third-generation neural networks, to adversarial examples is studied. The study considers rate and time encoding, as well as rate and first-to-spike decoding. Furthermore, a robust training mechanism is proposed that is demonstrated to enhance the performance of SNNs under white-box attacks.

Original language	English (US)
Title of host publication	2018 IEEE 19th International Workshop on Signal Processing Advances in Wireless Communications, SPAWC 2018
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Print)	9781538635124
DOIs	https://doi.org/10.1109/SPAWC.2018.8446003
State	Published - Aug 24 2018
Event	19th IEEE International Workshop on Signal Processing Advances in Wireless Communications, SPAWC 2018 - Kalamata, Greece Duration: Jun 25 2018 → Jun 28 2018

Publication series

Name	IEEE Workshop on Signal Processing Advances in Wireless Communications, SPAWC
Volume	2018-June

Other

Other	19th IEEE International Workshop on Signal Processing Advances in Wireless Communications, SPAWC 2018
Country/Territory	Greece
City	Kalamata
Period	6/25/18 → 6/28/18

All Science Journal Classification (ASJC) codes

Electrical and Electronic Engineering
Computer Science Applications
Information Systems

Keywords

Generalized Linear Model (GLM)
Spiking Neural Networks (SNNs)
adversarial examples
adversarial training

Access to Document

10.1109/SPAWC.2018.8446003

Cite this

Bagheri, A., Simeone, O., & Rajendran, B. (2018). Adversarial Training for Probabilistic Spiking Neural Networks. In 2018 IEEE 19th International Workshop on Signal Processing Advances in Wireless Communications, SPAWC 2018 Article 8446003 (IEEE Workshop on Signal Processing Advances in Wireless Communications, SPAWC; Vol. 2018-June). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/SPAWC.2018.8446003

Bagheri, Alireza ; Simeone, Osvaldo ; Rajendran, Bipin. / Adversarial Training for Probabilistic Spiking Neural Networks. 2018 IEEE 19th International Workshop on Signal Processing Advances in Wireless Communications, SPAWC 2018. Institute of Electrical and Electronics Engineers Inc., 2018. (IEEE Workshop on Signal Processing Advances in Wireless Communications, SPAWC).

@inproceedings{b6a948a963b14864a6afb931986d9b1a,

title = "Adversarial Training for Probabilistic Spiking Neural Networks",

abstract = "Classifiers trained using conventional empirical risk minimization or maximum likelihood methods are known to suffer dramatic performance degradations when tested over examples adversarially selected based on knowledge of the clas-sifier's decision rule. Due to the prominence of Artificial Neural Networks (ANNs) as classifiers, their sensitivity to adversarial examples, as well as robust training schemes, have been recently the subject of intense investigation. In this paper, for the first time, the sensitivity of spiking neural networks (SNNs), or third-generation neural networks, to adversarial examples is studied. The study considers rate and time encoding, as well as rate and first-to-spike decoding. Furthermore, a robust training mechanism is proposed that is demonstrated to enhance the performance of SNNs under white-box attacks.",

keywords = "Generalized Linear Model (GLM), Spiking Neural Networks (SNNs), adversarial examples, adversarial training",

author = "Alireza Bagheri and Osvaldo Simeone and Bipin Rajendran",

note = "Publisher Copyright: {\textcopyright} 2018 IEEE.; 19th IEEE International Workshop on Signal Processing Advances in Wireless Communications, SPAWC 2018 ; Conference date: 25-06-2018 Through 28-06-2018",

year = "2018",

month = aug,

day = "24",

doi = "10.1109/SPAWC.2018.8446003",

language = "English (US)",

isbn = "9781538635124",

series = "IEEE Workshop on Signal Processing Advances in Wireless Communications, SPAWC",

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

booktitle = "2018 IEEE 19th International Workshop on Signal Processing Advances in Wireless Communications, SPAWC 2018",

address = "United States",

}

Bagheri, A, Simeone, O & Rajendran, B 2018, Adversarial Training for Probabilistic Spiking Neural Networks. in 2018 IEEE 19th International Workshop on Signal Processing Advances in Wireless Communications, SPAWC 2018., 8446003, IEEE Workshop on Signal Processing Advances in Wireless Communications, SPAWC, vol. 2018-June, Institute of Electrical and Electronics Engineers Inc., 19th IEEE International Workshop on Signal Processing Advances in Wireless Communications, SPAWC 2018, Kalamata, Greece, 6/25/18. https://doi.org/10.1109/SPAWC.2018.8446003

Adversarial Training for Probabilistic Spiking Neural Networks. / Bagheri, Alireza; Simeone, Osvaldo; Rajendran, Bipin.
2018 IEEE 19th International Workshop on Signal Processing Advances in Wireless Communications, SPAWC 2018. Institute of Electrical and Electronics Engineers Inc., 2018. 8446003 (IEEE Workshop on Signal Processing Advances in Wireless Communications, SPAWC; Vol. 2018-June).

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

TY - GEN

T1 - Adversarial Training for Probabilistic Spiking Neural Networks

AU - Bagheri, Alireza

AU - Simeone, Osvaldo

AU - Rajendran, Bipin

PY - 2018/8/24

Y1 - 2018/8/24

N2 - Classifiers trained using conventional empirical risk minimization or maximum likelihood methods are known to suffer dramatic performance degradations when tested over examples adversarially selected based on knowledge of the clas-sifier's decision rule. Due to the prominence of Artificial Neural Networks (ANNs) as classifiers, their sensitivity to adversarial examples, as well as robust training schemes, have been recently the subject of intense investigation. In this paper, for the first time, the sensitivity of spiking neural networks (SNNs), or third-generation neural networks, to adversarial examples is studied. The study considers rate and time encoding, as well as rate and first-to-spike decoding. Furthermore, a robust training mechanism is proposed that is demonstrated to enhance the performance of SNNs under white-box attacks.

AB - Classifiers trained using conventional empirical risk minimization or maximum likelihood methods are known to suffer dramatic performance degradations when tested over examples adversarially selected based on knowledge of the clas-sifier's decision rule. Due to the prominence of Artificial Neural Networks (ANNs) as classifiers, their sensitivity to adversarial examples, as well as robust training schemes, have been recently the subject of intense investigation. In this paper, for the first time, the sensitivity of spiking neural networks (SNNs), or third-generation neural networks, to adversarial examples is studied. The study considers rate and time encoding, as well as rate and first-to-spike decoding. Furthermore, a robust training mechanism is proposed that is demonstrated to enhance the performance of SNNs under white-box attacks.

KW - Generalized Linear Model (GLM)

KW - Spiking Neural Networks (SNNs)

KW - adversarial examples

KW - adversarial training

UR - http://www.scopus.com/inward/record.url?scp=85053456439&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85053456439&partnerID=8YFLogxK

U2 - 10.1109/SPAWC.2018.8446003

DO - 10.1109/SPAWC.2018.8446003

M3 - Conference contribution

AN - SCOPUS:85053456439

SN - 9781538635124

T3 - IEEE Workshop on Signal Processing Advances in Wireless Communications, SPAWC

BT - 2018 IEEE 19th International Workshop on Signal Processing Advances in Wireless Communications, SPAWC 2018

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 19th IEEE International Workshop on Signal Processing Advances in Wireless Communications, SPAWC 2018

Y2 - 25 June 2018 through 28 June 2018

ER -

Bagheri A, Simeone O, Rajendran B. Adversarial Training for Probabilistic Spiking Neural Networks. In 2018 IEEE 19th International Workshop on Signal Processing Advances in Wireless Communications, SPAWC 2018. Institute of Electrical and Electronics Engineers Inc. 2018. 8446003. (IEEE Workshop on Signal Processing Advances in Wireless Communications, SPAWC). doi: 10.1109/SPAWC.2018.8446003

Adversarial Training for Probabilistic Spiking Neural Networks

Abstract

Publication series

Other

All Science Journal Classification (ASJC) codes

Keywords

Access to Document

Other files and links

Fingerprint

Cite this