Adversarial Examples: Opportunities and Challenges

TL;DR

This paper reviews the latest research on adversarial examples in deep neural networks, discussing their creation, detection, limitations, and future challenges in AI security applications.

Contribution

It provides a comprehensive survey of recent advances in adversarial example generation and defense methods, highlighting current limitations and future research directions.

Findings

Adversarial examples can mislead DNNs without human detection.

Current defense methods have notable limitations.

Future research needs to address robustness and detection challenges.

Abstract

Deep neural networks (DNNs) have shown huge superiority over humans in image recognition, speech processing, autonomous vehicles and medical diagnosis. However, recent studies indicate that DNNs are vulnerable to adversarial examples (AEs), which are designed by attackers to fool deep learning models. Different from real examples, AEs can mislead the model to predict incorrect outputs while hardly be distinguished by human eyes, therefore threaten security-critical deep-learning applications. In recent years, the generation and defense of AEs have become a research hotspot in the field of artificial intelligence (AI) security. This article reviews the latest research progress of AEs. First, we introduce the concept, cause, characteristics and evaluation metrics of AEs, then give a survey on the state-of-the-art AE generation methods with the discussion of advantages and disadvantages. After that, we review the existing defenses and discuss their limitations. Finally, future research opportunities and challenges on AEs are prospected.