Attacking a CNN-based Layout Hotspot Detector Using Group Gradient Method

Haoyu Yang; Shifan Zhang; Kang Liu; Siting Liu; Benjamin Tan; Ramesh Karri; Siddharth Garg; Bei Yu; Evangeline F.Y. Young

doi:10.1145/3394885.3431571

Attacking a CNN-based Layout Hotspot Detector Using Group Gradient Method

Haoyu Yang, Shifan Zhang, Kang Liu, Siting Liu, Benjamin Tan, Ramesh Karri, Siddharth Garg, Bei Yu, Evangeline F.Y. Young

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

Abstract

Deep neural networks are being used in disparate VLSI design automation tasks, including layout printability estimation, mask optimization, and routing congestion analysis. Preliminary results show the power of deep learning as an alternate solution in state-of-theart design and sign-off flows. However, deep learning is vulnerable to adversarial attacks. In this paper, we examine the risk of state-ofthe- art deep learning-based layout hotspot detectors under practical attack scenarios. We show that legacy gradient-based attacks do not adequately consider the design rule constraints. We present an innovative adversarial attack formulation to attack the layout clips and propose a fast group gradient method to solve it. Experiments show that the attack can deceive the deep neural networks using small perturbations in clips which preserve layout functionality while meeting the design rules. The source code is available at https://github.com/phdyang007/dlhsd/tree/dct_as_conv.

Original language	English (US)
Title of host publication	Proceedings of the 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	885-891
Number of pages	7
ISBN (Electronic)	9781450379991
DOIs	https://doi.org/10.1145/3394885.3431571
State	Published - Jan 18 2021
Event	26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021 - Virtual, Online, Japan Duration: Jan 18 2021 → Jan 21 2021

Publication series

Name	Proceedings of the Asia and South Pacific Design Automation Conference, ASP-DAC

Conference

Conference	26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021
Country/Territory	Japan
City	Virtual, Online
Period	1/18/21 → 1/21/21

ASJC Scopus subject areas

Electrical and Electronic Engineering
Computer Science Applications
Computer Graphics and Computer-Aided Design

Access to Document

10.1145/3394885.3431571

Cite this

Yang, H., Zhang, S., Liu, K., Liu, S., Tan, B., Karri, R., Garg, S., Yu, B., & Young, E. F. Y. (2021). Attacking a CNN-based Layout Hotspot Detector Using Group Gradient Method. In Proceedings of the 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021 (pp. 885-891). Article 3431571 (Proceedings of the Asia and South Pacific Design Automation Conference, ASP-DAC). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1145/3394885.3431571

Attacking a CNN-based Layout Hotspot Detector Using Group Gradient Method. / Yang, Haoyu; Zhang, Shifan; Liu, Kang et al.
Proceedings of the 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021. Institute of Electrical and Electronics Engineers Inc., 2021. p. 885-891 3431571 (Proceedings of the Asia and South Pacific Design Automation Conference, ASP-DAC).

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

Yang, H, Zhang, S, Liu, K, Liu, S, Tan, B, Karri, R , Garg, S, Yu, B & Young, EFY 2021, Attacking a CNN-based Layout Hotspot Detector Using Group Gradient Method. in Proceedings of the 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021., 3431571, Proceedings of the Asia and South Pacific Design Automation Conference, ASP-DAC, Institute of Electrical and Electronics Engineers Inc., pp. 885-891, 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021, Virtual, Online, Japan, 1/18/21. https://doi.org/10.1145/3394885.3431571

Yang H, Zhang S, Liu K, Liu S, Tan B, Karri R et al. Attacking a CNN-based Layout Hotspot Detector Using Group Gradient Method. In Proceedings of the 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021. Institute of Electrical and Electronics Engineers Inc. 2021. p. 885-891. 3431571. (Proceedings of the Asia and South Pacific Design Automation Conference, ASP-DAC). doi: 10.1145/3394885.3431571

Yang, Haoyu ; Zhang, Shifan ; Liu, Kang et al. / Attacking a CNN-based Layout Hotspot Detector Using Group Gradient Method. Proceedings of the 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021. Institute of Electrical and Electronics Engineers Inc., 2021. pp. 885-891 (Proceedings of the Asia and South Pacific Design Automation Conference, ASP-DAC).

@inproceedings{c4141206db8043818c6c09821ca7fa89,

title = "Attacking a CNN-based Layout Hotspot Detector Using Group Gradient Method",

abstract = "Deep neural networks are being used in disparate VLSI design automation tasks, including layout printability estimation, mask optimization, and routing congestion analysis. Preliminary results show the power of deep learning as an alternate solution in state-of-theart design and sign-off flows. However, deep learning is vulnerable to adversarial attacks. In this paper, we examine the risk of state-ofthe- art deep learning-based layout hotspot detectors under practical attack scenarios. We show that legacy gradient-based attacks do not adequately consider the design rule constraints. We present an innovative adversarial attack formulation to attack the layout clips and propose a fast group gradient method to solve it. Experiments show that the attack can deceive the deep neural networks using small perturbations in clips which preserve layout functionality while meeting the design rules. The source code is available at https://github.com/phdyang007/dlhsd/tree/dct_as_conv.",

author = "Haoyu Yang and Shifan Zhang and Kang Liu and Siting Liu and Benjamin Tan and Ramesh Karri and Siddharth Garg and Bei Yu and Young, {Evangeline F.Y.}",

note = "Publisher Copyright: {\textcopyright} 2021 Association for Computing Machinery.; 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021 ; Conference date: 18-01-2021 Through 21-01-2021",

year = "2021",

month = jan,

day = "18",

doi = "10.1145/3394885.3431571",

language = "English (US)",

series = "Proceedings of the Asia and South Pacific Design Automation Conference, ASP-DAC",

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

pages = "885--891",

booktitle = "Proceedings of the 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021",

}

TY - GEN

T1 - Attacking a CNN-based Layout Hotspot Detector Using Group Gradient Method

AU - Yang, Haoyu

AU - Zhang, Shifan

AU - Liu, Kang

AU - Liu, Siting

AU - Tan, Benjamin

AU - Karri, Ramesh

AU - Garg, Siddharth

AU - Yu, Bei

AU - Young, Evangeline F.Y.

PY - 2021/1/18

Y1 - 2021/1/18

N2 - Deep neural networks are being used in disparate VLSI design automation tasks, including layout printability estimation, mask optimization, and routing congestion analysis. Preliminary results show the power of deep learning as an alternate solution in state-of-theart design and sign-off flows. However, deep learning is vulnerable to adversarial attacks. In this paper, we examine the risk of state-ofthe- art deep learning-based layout hotspot detectors under practical attack scenarios. We show that legacy gradient-based attacks do not adequately consider the design rule constraints. We present an innovative adversarial attack formulation to attack the layout clips and propose a fast group gradient method to solve it. Experiments show that the attack can deceive the deep neural networks using small perturbations in clips which preserve layout functionality while meeting the design rules. The source code is available at https://github.com/phdyang007/dlhsd/tree/dct_as_conv.

AB - Deep neural networks are being used in disparate VLSI design automation tasks, including layout printability estimation, mask optimization, and routing congestion analysis. Preliminary results show the power of deep learning as an alternate solution in state-of-theart design and sign-off flows. However, deep learning is vulnerable to adversarial attacks. In this paper, we examine the risk of state-ofthe- art deep learning-based layout hotspot detectors under practical attack scenarios. We show that legacy gradient-based attacks do not adequately consider the design rule constraints. We present an innovative adversarial attack formulation to attack the layout clips and propose a fast group gradient method to solve it. Experiments show that the attack can deceive the deep neural networks using small perturbations in clips which preserve layout functionality while meeting the design rules. The source code is available at https://github.com/phdyang007/dlhsd/tree/dct_as_conv.

UR - http://www.scopus.com/inward/record.url?scp=85100597118&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85100597118&partnerID=8YFLogxK

U2 - 10.1145/3394885.3431571

DO - 10.1145/3394885.3431571

M3 - Conference contribution

AN - SCOPUS:85100597118

T3 - Proceedings of the Asia and South Pacific Design Automation Conference, ASP-DAC

SP - 885

EP - 891

BT - Proceedings of the 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021

Y2 - 18 January 2021 through 21 January 2021

ER -

Attacking a CNN-based Layout Hotspot Detector Using Group Gradient Method

Abstract

Publication series

Conference

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this