Search papers, labs, and topics across Lattice.
This paper explores the use of Transformer-based Language Models (TLMs) for detecting anomalies in HTTP requests, addressing the challenges posed by mislabeled or contaminated training data in semi-supervised learning environments. By employing token-level logit-based surprisal mapping, the authors provide both an anomaly score and detailed explanations for detected anomalies, enhancing the interpretability of the detection process. The findings reveal significant labeling inconsistencies in a widely used dataset, highlighting how such contamination can lead to backdoor-like failures in anomaly detection models.
Anomalous contamination in training data can silently compromise web anomaly detection, leading to backdoor failures that are hard to detect.
The rapid growth of web-based services, particularly API-driven architectures, reflects an increasing reliance on distributed systems, exposing sensitive data to security risks and making the adoption of automated defensive mechanisms essential. In this context, where benign traffic predominates in real-world settings, modern defenses increasingly model normal behavior, relying on semi-supervised approaches trained on only normal data. However, ensuring the complete absence of anomalous instances in such training data is inherently difficult in practice, and mislabeled or contaminated attack samples can introduce backdoors into the learned defense, causing the model to silently misclassify certain attack patterns as normal behavior. This paper investigates the effectiveness of Transformer-based Language Models (TLMs) in the detection of anomalies in HTTP requests, focussing on providing detailed explanations for the detected anomalies. The study employs token-level logit-based surprisal mapping to provide both an anomaly score and a direct, detailed explanation via heatmap-like highlighting. The effectiveness of the proposed explainability approach is demonstrated by the discovery of labelling inconsistencies in a popular public dataset, revealing how anomalous contamination in the training data had induced backdoor-like failures in the detection models.
