How moda-gov-tw/pets-applications works
Project type and purpose
• This is a multi-module research/demo repository showcasing six Privacy-Enhancing Technology (PET) applications:
• Differential Privacy
• Synthetic Data
• Federated Learning
• Homomorphic Encryption
• Secure Multiparty Computation
• k-Anonymization
• Each top-level folder implements one PET use case (mostly Python scripts or small apps) to demonstrate data-sharing or compute workflows while preserving privacy.
File structure
• Top-level directories correspond to PET categories. Inside each:
• A README.md describing the use case
• Sample datasets (NHANES.csv, other CSVs)
• Example scripts (.py, .sh)
• Subfolders for configurations, code, figures, utilities
• The federated_learning/financial_fraud_detection module has a richer structure:
– Shell scripts to generate job configs and splits
– A “tree-based” submodule with client/server JSON configs, custom executors, meta.json, and TensorBoard utilities
• The homomorphic_encryption module contains two distinct apps:
– A Node.js web frontend (views, assets, main.js) plus server code (TTC.js)
– A C++ TFHE-based compute library with CMake, include/, src/, testing/
• The secure_multiparty_computation module implements a small multi-party script hierarchy (Client, MPC_node, Player-Data).
• Config files (package.json, requirements.txt, CMakeLists.txt) signal different runtime environments (Node.js, Python, C++).
README for additional insights
• Confirms the high-level grouping and purpose of each directory
• Disclaimer: demo code only