Many applications have used machine learning as a tool to detect malware. These applications take in raw or processed binary data to feed neural network models to classify benign or malicious files. Even though this approach has proved effective against dynamic changes, such as encrypting, obfuscating and packing techniques, it is vulnerable to specific evasion attacks to where that small changes to the input data cause misclassification at test time. In this paper, I propose MDEA, an Adversarial Malware Detection model that combines a neural network and evolutionary optimization attack samples to make the network robust against evasion attacks. By retraining the model with the evolved malware samples, network performance improves a big margin.

PDF

Masters Thesis, Department of Computer Science, The University of Texas at Austin, 2019.

@mastersthesis{wang:thesis2019, title={MDEA: Malware Detection with Evolutionary Adversarial Learning }, author={Xiruo Wang}, month={ }, school={Department of Computer Science, The University of Texas at Austin}, url="http://nn.cs.utexas.edu/?wang:thesis2019", year={2019} }