So has there been any success in using deep learning for doing reverse engineering?
i couldn't find anything useful other than some theoretical papers, so was there any talk about this in recent security conferences, or actual success of doing so?
Asked
Active
Viewed 298 times
2
-
1Interesting topic. What do you want exactly to reach with deep learning? The disassembly process? Or classification of malware? Recognize obfuscation? – pudi Oct 08 '18 at 21:51
-
Are you focused just on deep learning specifically or machine learning methods in RE in general? – julian Oct 08 '18 at 23:40
-
@pudi yes, building a "program" with the help of deep learning that can disassemble any given program with a high accuracy rate, has there been any success in this field? – John P Oct 09 '18 at 05:49
-
@SYS_V deep learning would be preferable – John P Oct 09 '18 at 07:59
-
6Possible duplicate of Can AI be used to write better decompilers/disassemblers? – perror Oct 09 '18 at 11:54
1 Answers
3
I don't know about deep learning, but if you are looking for general results on using machine learning for reverse engineering, there was a paper called "Evolving Exact Decompilation" published at the Workshop on Binary Analysis Research 2018, where the researchers claimed to have learned/evolved the ability to decompile a C program. See the paper here.
dingo_kinznerhook
- 1,016
- 8
- 34