Human-Computer Decision Systems for Cybersecurity
November 2016 • Presentation
This work discovered a surprising result regarding the potential for non-experts to perform malware family analsys
Software Engineering Institute
In this work, we studied multiple facts of human-ML collaboration, using both real malware classification problems and a model problem based on malware classification. We investigated methods using both supervised (active) and unsupervised learning to augment the abilities of analysts. We also discovered a surprising result regarding the potential for nonexperts to perform malware family analysis using low-imensional visualizations.