Implementation of Differentially Private Partitioned Variational Inference. The implementation is based on the original PVI code base.

Install dependencies in notebooks/examples/requirement.txt (pip) and environment.yml (Conda/Mamba). Download and preprocess required data sets.

All codes for running experiments are located in notebooks/examples.

To run logistic regression model experiments, use 'dp_logistic_regression.py' after checking the argparser arguments. For BNN, use instead 'run_bnn.py'.