Efficiency Robustness of Dynamic Deep Learning Systems

2025-06-12

AI Summary

This work addresses a novel efficiency robustness threat to Dynamic Deep Learning Systems (DDLSs) arising from their inherent dynamic mechanisms. We first establish a systematic taxonomy of efficiency-oriented adversarial attacks, categorizing them into three attack surfaces: dynamic computational cost, dynamic inference rounds, and dynamic output generation. Leveraging adversarial modeling, runtime behavioral monitoring, and system-level security evaluation, we expose the fundamental failure of existing defenses against such efficiency attacks and empirically demonstrate their widespread vulnerability across diverse DDLSs. Our study identifies the core challenges in achieving efficiency robustness for DDLSs and articulates the critical need for adaptive, defense-aware system design. The findings provide both theoretical foundations and methodological support for the trustworthy deployment of dynamic AI systems in resource-constrained environments.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Multiagent Systems: Adversarial Agents

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Deep Learning Systems (DLSs) are increasingly deployed in real-time applications, including those in resourceconstrained environments such as mobile and IoT devices. To address efficiency challenges, Dynamic Deep Learning Systems (DDLSs) adapt inference computation based on input complexity, reducing overhead. While this dynamic behavior improves efficiency, such behavior introduces new attack surfaces. In particular, efficiency adversarial attacks exploit these dynamic mechanisms to degrade system performance. This paper systematically explores efficiency robustness of DDLSs, presenting the first comprehensive taxonomy of efficiency attacks. We categorize these attacks based on three dynamic behaviors: (i) attacks on dynamic computations per inference, (ii) attacks on dynamic inference iterations, and (iii) attacks on dynamic output production for downstream tasks. Through an in-depth evaluation, we analyze adversarial strategies that target DDLSs efficiency and identify key challenges in securing these systems. In addition, we investigate existing defense mechanisms, demonstrating their limitations against increasingly popular efficiency attacks and the necessity for novel mitigation strategies to secure future adaptive DDLSs.

Problem

• Explores efficiency robustness in Dynamic Deep Learning Systems
• Analyzes adversarial attacks targeting dynamic computation mechanisms
• Identifies defense limitations and need for novel mitigation strategies

Innovation

• Dynamic computation adaptation for efficiency
• Taxonomy of efficiency adversarial attacks
• Evaluation of defense mechanism limitations