Graphic Content Ahead - Towards Automated Scalable Analysis of Graphical Images Embedded in Malware

Explore automated techniques for analyzing graphical images embedded in malware in this 25-minute Black Hat conference talk. Delve into two key problems: identifying malware samples with visually similar image sets and classifying malware images into topical categories. Learn about a scale and contrast invariant approach for reducing images to low-dimensional binary vectors, indexing techniques for approximating Hamming distance, and force-directed graph visualization for displaying results. Discover how to dynamically obtain labeled training examples using the Google Image Search API and compare various image classifiers for categorizing malware images. Gain insights into the effectiveness of these techniques for different classes of malware images and understand the potential impact on malware triage and attribution processes.

Graphic Content Ahead: Towards Automated Scalable Analysis Of Graphical Images Embedded In Malware