A Foray Into Extracting Malicious Features from Executable Code with Neural Network Salience
Michael Santacroce, Wayne Stegner, Daniel Koranek, Rashmi Jha · 2019
We have previously created successful neural networks for malware detection. Here, we examine a network with salience to extract parts of an input deemed important. We show that the blocks we extract are what is important to the network, are unique to their class, and show clear similarities when clustered.