Projects

Malware Family Classification

  • Implemented Kernel Subspace Method (KSM) for malware classification based on representative digital images. [code] [paper]
  • Results published in IEEE access.

Malware Spoofing (Binary)

  • Developed a method based on Kernel Constrained Subspace Method (KCSM) and Random Fourier Features (RFF). [code] [paper]
  • Tackled malware spoofing task, distinguishing between malware and Benign (safe) files.
  • Results published in MVA2023.

Malware Visualization (Behavior)

  • Proposed novel visualization framework (OSA_KDS) for understanding malware behavior. [code] [paper]
  • Framework implemented using Occlusion Sensitivity Analysis and Kernel Difference Subspace.
  • Provided insights into malware features, aiding in reverse-engineering and cybersecurity analysis.
  • Results published on IEEE access.