Proof of concept implementation of a Privug backend featuring an exact Bayesian inference engine based on multivariate Gaussian distributions.
This repository accompanies the paper "Exact and Efficient Bayesian Inference for Privacy Risk Quantification".
This repository contains the public statistics release case study and the benchmarks for the scalability evaluation of the inference engine. The examples can be executed in the notebook eval.ipynb. The folder case_study_files contains the files for the case study. The scalability folder scalability_files contains the template programs for the scalability evaluation. The evaluation functions in utils/eval.py instantiate the benchmark programs with increasing number of variables.