This is a machine learning project for vulnerability patches.
- I- Introduction VulnCatcher
- II- Dataset used
- III-Environment
- IV-How to run
- V- Results: Detected security patches examples
- VI- Results: Confusion matrix
- VII- State of art comparison
VulnCatcher a semi-supervised learning approach for security patches detection.
This project is based on Data7 Tool(https://github.com/electricalwind/data7) for Labeled examples and each fix commits for unlabeled examples.
Compile files with : python setup.py
- Download Python 2.7
- Compile files with : python setup.py
To run this tool, please check instructions on setup.py.
We chose some examples on prediction set to check if our approach really detects security patches.
- Linux project examples
- commit 61656dd2e62f91b194b803f15c6faf0a647dcdf9
- commit 724519d8f987b069867cb9b0cf25a50116402f37
- Wireshark project examples
- commit 6b13c05da11e7735b4a50995c23ecf309d55a62d
- commit 8c959c80e983f0500b7abd4d73d0b6e845e941c0
- OpenSSL project examples
- commit d0c98589146d79f1059638057dad9bb80d662339
- commit 9ee1c838cbfeb1571979198ca6891a539ae3d931