Exploring DNN Robustness Against Adversarial Attacks Using Approximate Multipliers

Deep Neural Networks (DNNs) have advanced in many real-world applications, such as healthcare and autonomous driving. However, their high computational complexity and vulnerability to adversarial attacks are ongoing challenges. In this letter, approximate multipliers are used to explore DNN robustne...

Full description

Saved in:
Bibliographic Details
Main Authors Askarizadeh, Mohammad Javad, Farahmand, Ebrahim, Castro-Godinez, Jorge, Mahani, Ali, Cabrera-Quiros, Laura, Salazar-Garcia, Carlos
Format Journal Article
LanguageEnglish
Published 17.04.2024
Subjects
Computer Science - Cryptography and Security
Computer Science - Learning
Online AccessGet full text

Cover

More Information
Summary:Deep Neural Networks (DNNs) have advanced in many real-world applications, such as healthcare and autonomous driving. However, their high computational complexity and vulnerability to adversarial attacks are ongoing challenges. In this letter, approximate multipliers are used to explore DNN robustness improvement against adversarial attacks. By uniformly replacing accurate multipliers for state-of-the-art approximate ones in DNN layer models, we explore the DNNs robustness against various adversarial attacks in a feasible time. Results show up to 7% accuracy drop due to approximations when no attack is present while improving robust accuracy up to 10% when attacks applied.
DOI:10.48550/arxiv.2404.11665