Skip to content

Conversation

@T-Breezy444
Copy link

Add Data-Free Attack Implementation

Changes Overview

New Directory Structure

Added new data_free_attack/ directory containing:

data_free_attack/
├── attacks/
│ ├── attack1.py
│ ├── attack2.py
│ └── attack3.py
├── models/
│ ├── generator.py
│ ├── surrogate.py
│ └── victim.py
├── readme.md
└── example.py

Dependency Updates

Version Adjustments

  • Changed numpy==2.0.1 to numpy>=1.23.5,<2.0.0 to resolve conflicts with matplotlib and other dependencies
  • Updated environment.yml to reflect the new numpy version requirements

New Required Packages

Added the following dependencies to requirements.txt:

  • torch-geometric==2.5.0
  • ogb==1.3.6
  • dgl==2.2.1
  • scikit-learn==1.3.0
  • matplotlib==3.7.5

README.md

Updated installation documentation:

  • Added support instructions for CUDA 11.x and 12.x
  • Improved troubleshooting steps based on cross-device testing
  • Added verification steps for CUDA setup
  • Included both direct and environment-based installation methods

Purpose

These additions implement data-free attacks functionality within the library. The new modules provide a complete framework for executing and evaluating data-free attacks.

Contact

For questions or clarifications about this PR, please contact me, I am more than happy to help!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant