Explainable ML‐Based Detection of Network‐Based Fuzzing Attacks

The concept of fuzzing was mainly used in software testing to ensure that software products were reliable and resilient to varying inputs. However, malicious actors adopted fuzzing as a method of pushing software, and mainly web applications, to misbehave under rapid-fire style random inputs. This chapter presents a machine learning (ML)-based fuzzing detection system that inspects network traffic to detect fuzzing. The proposed system uses flow-based network traffic analysis to identify malicious fuzzing attacks using ML classifiers. The chapter reviews relevant studies in fuzzing detection and ML-based network intrusion detection systems. It introduces the proposed detection framework and details the experimental setup. The chapter also presents the results and analysis, discussing the key findings. Finally, it outlines directions for future work.