Understanding Honeypot Data by an Unsupervised Neural Visualization

Abstract

Neural projection techniques can adaptively map high-dimensional data into a low-dimensional space, for the user-friendly visualization of data collected by different security tools. Such techniques are applied in this study for the visual inspection of honeypot data, which may be seen as a complementary network security tool that sheds light on internal data structures through visual inspection. Empirical verification of the proposed projection methods was performed in an experimental domain where data were captured from a honeypot network. Experiments showed that visual inspection of these data, contributes to easily gain a deep understanding of attack patterns and strategies.