Designing Security-Enhanced Architectures for Analog Neural Networks

##plugins.themes.academic_pro.article.main##

Hadia Azmat
Zilly Huma

Abstract

Analog neural networks (ANNs) are gaining attention for their potential in achieving energy efficiency and low-latency computations compared to their digital counterparts. However, the security of ANNs remains a significant challenge due to vulnerabilities such as adversarial attacks, data leakage, and hardware-level manipulations. This paper explores the design principles and strategies for enhancing the security of ANNs. By examining the unique characteristics of analog systems, the paper identifies key vulnerabilities and proposes novel architectural solutions. The results demonstrate that incorporating security measures during the design phase can significantly improve the robustness and reliability of ANNs, paving the way for their broader adoption in critical applications.

##plugins.themes.academic_pro.article.details##

How to Cite
Hadia Azmat, & Zilly Huma. (2025). Designing Security-Enhanced Architectures for Analog Neural Networks. Pioneer Research Journal of Computing Science, 1(2), 1–6. Retrieved from http://prjcs.com/index.php/prjcs/article/view/9