In this project, I implemented a logistic regression model capable of operating on an encrypted dataset using homomorphic encryption. This approach ensures data security while performing computations on sensitive data, addressing significant privacy concerns. The project involves designing algorithms, preprocessing data, and evaluating model performance while keeping the data encrypted throughout the process.
- Design Algorithms: Developed and designed the logistic regression algorithm tailored to work with encrypted data.
- Data Exploration and Preprocessing: Explored and preprocessed dataset to prepare it for encryption.
- Training and Evaluation: Train logistic regression models and evaluate their performance to compare average accuracy and AUC score.
- Python
- Jupyter Notebook
- Microsoft SEAL
- NumPy & Pandas