Visual-Interactive System Explains Network Traffic Predictions with Aggregated Class Activation Maps

A new research paper on arXiv (2608.13575) introduces a visual-interactive system designed to explain deep learning model predictions for network traffic classification. The system uses aggregated class activation maps to provide global explanations derived from multiple samples of a given class, addressing the challenge of diverging patterns within a single predicted class. This tool is crucial for network traffic analysis and intrusion detection, particularly when using advanced tools like next-generation firewalls. The paper emphasizes the need for descriptive rules for classes and the ability to detect and analyze patterns within predicted classes. The system aims to enhance the interpretability of machine learning models in the context of computer network traffic classification, a domain where deep learning has shown impressive results. The research highlights the importance of clear and comprehensive explanations in network security applications.