Asm2Vec: boosting static representation robustness for binary clone search against code obfuscation


Binary analysis is one of the critical processes
for malware analysis, vulnerability studies, and software plagiarism detection. However,
this process is manually intensive. A binary can also be obfuscated or optimized into different
forms which pose great challenges to this process. Kam1n0 is an open-source binary analysis and
assembly code management platform. It enables the user to create multiple repositories and
index different binaries. Given a binary file under analysis, it can efficiently search
for any clones of its assembly functions, even for the highly optimized or obfuscated
assembly code. So how do we do it? Kam1n0 builds a neural
network named Asm2Vec. It reads a large volume of assembly code, and tries to understand
their underlying latent dimensions. Given any code fragment, it converts it into a numeric
vector that represents its uniqueness for clone search. Check out Kam1n0 today on GitHub.

Leave a Reply

Your email address will not be published. Required fields are marked *