23ss-BinaryML

Practical "AI Methods and Tools for Programming"

BinaryML : Classifying Binaries for Malware and Vulnerability Detection

The main goal of this project is to combine and improve best machine learning approaches for detection of malware and vulnerabilities based on binaries.

Objectives

1. Comprehensive Literature Review and Comparison.
2. Identification and Prioritization of Best Approaches.
3. Cross domain code transfer: Bridging the Gap between Vulnerability and Malware Detection.
4. Evaluation on Real Binaries and Benchmarking!

Two security detection models for malware and vulnerability domains make up the majority of this project. For more information and setup, see the readme files in the respective directory.

1. Malware detection model: MalwareDetectionHRR/
2. Vulnerability detection model: VulnerabilityDetectionRomeo/

Below is the link to the results of our literature search which enabled us to identify the best apporach in each domain: https://docs.google.com/spreadsheets/d/18S8JuEuAODwQkR4lEggQR33j6HUOi8ETXkKnT6neS9A/edit#gid=0

You can find the detailed developer's documentation, final presentation PPT, and other docs related to the project in docs/ directory.