Malware Detection in Executables Using Neural Networks

Originally published at: https://developer.nvidia.com/blog/malware-detection-neural-networks/

The detection of malicious software (malware) is an increasingly important cyber security problem for all of society. Single incidences of malware can cause millions of dollars in damage. The current generation of anti-virus and malware detection products typically use a signature-based approach, where a set of manually crafted rules attempt to identify different groups of known…

The current generation of anti-virus and malware detection products typically use a signature-based approach, where a set of manually crafted rules attempt to identify different groups of known malware types. These rules are generally specific and brittle, and usually unable to recognize new malware even if it uses the same functionality.

Indeed. At the risk of sounding like a company shill, this is exactly why Cylance was founded 5 years ago. You can read more about Cylance Protect here.

Good article. Whether the data and code of the experiments is available for further research.

Excuse me, I have a question about the training data. In the article, you imply that the size of an exe is almost 1-2M. But usually, more exes have a size more than that. If the size is bigger, the memery and calculation problems will appear. How can we reduce the size of an exe to 1-2M?

With simple crop would be enought to resolve your problem.