Magical-Decomposition: Winning Both Adversarial Robustness and Efficiency on Hardware

Model compression is one of the most preferred techniques for efficiently deploying deep neural networks (DNNs) on resource- constrained Internet of Things (IoT) platforms. However, the simply compressed model is often vulnerable to adversarial attacks, leading to a conflict between robustness and e...

Full description

Saved in:
Bibliographic Details
Main Authors: Cheng, Xin, Wang, Mei-Qi, Shi, Yu-Bo, Lin, Jun, Wang, Zhong-Feng
Format: Conference Proceeding
Language:English
Subjects:
Adversarial training
Decomposition
Deep learning
Hardware acceleration
Internet of Things
Machine learning
Machine learning algorithms
Model compression
Neural networks
Robustness
Scalability
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Model compression is one of the most preferred techniques for efficiently deploying deep neural networks (DNNs) on resource- constrained Internet of Things (IoT) platforms. However, the simply compressed model is often vulnerable to adversarial attacks, leading to a conflict between robustness and efficiency, especially for IoT devices exposed to complex real-world scenarios. We, for the first time, address this problem by developing a novel framework dubbed Magical-Decomposition to simultaneously enhance both robustness and efficiency for hardware. By leveraging a hardware-friendly model compression method called singular value decomposition, the defending algorithm can be supported by most of the existing DNN hardware accelerators. To step further, by using a recently developed DNN interpretation tool, the underlying scheme of how the adversarial accuracy can be increased in the compressed model is highlighted clearly. Ablation studies and extensive experiments under various attacks/models/datasets consistently validate the effectiveness and scalability of the proposed framework.
ISSN:2160-1348
DOI:10.1109/ICMLC56445.2022.9941335