PhD Proposal: DNS Anomaly Investigation and Visualization using Immersive Data Analytics
The recent increase in popularity and technical realization of virtual reality makes possible new paradigms of visualization and interaction. In addition, deep learning has come to the forefront as a powerhouse of analytics and machine learning. In this proposal, we review our work on virtual reality visualization, cyber sickness mitigation, and deep-learning-driven data exploration and visualization, followed by our plans for a project that builds on that work to tackle the investigation and visualization of DNS attacks. We first present our work comparing the benefits of increased immersion on recall between head-mounted displays and traditional desktop displays. After conducting a user study where participants memorize and recall a series of spatially-distributed faces on both a desktop and head-mounted display, we found there was a statistically significant increase in recall accuracy of the names of the faces in the head-mounted display. Next, we present our work on measuring and quantifying cyber sickness through EEG analysis. We found that self-reported cyber sickness had statistically significant correlations with increases in delta and theta brain waves. This finding allows future virtual reality developers to be able to measure and quantify cyber sickness, and therefore design countermeasures to mitigate it. Lastly, we present our work on enabling the discovery of hidden labels and communities within high-dimensional datasets using visualization and deep learning. Our plans are to leverage the advances and discoveries made from our previous works to facilitate the visualization, interaction, and discovery of anomalies and attacks targeting a core pillar of the internet, the Domain Name System (DNS). Specifically, our goal is to develop a technique that allows for the discovery and analysis of nuances and patterns lurking within DNS attacks using virtual reality visualization and human interaction with analysis driven and supported by deep learning.
Chair: Dr. Amitabh Varsney
Dept rep: Dr. John Dickerson
Members: Dr. Matthias Zwicker
Dr. Joseph JaJa