Build a histogram for the adult dataset from UCI repository(http://archive.ics.uci.edu/ml/machine-learning-databases/adult/) Then, build a noisy histogram (differentially private) by adding random noise from Laplacian distribution and plot mean squared error for different values of epsilon (privacy budget).
Part 1. Synthetic Data Generation
"Synthetic database" is used in an interactive setting instead of original database. Synthetic database is constructed by learning the joint distribution of the data. This project assignment is for learning distribution by building a multi-dimensional histogram and privatize it using Laplace mechanism. For this task, we use the Tic-Tac-Toe game dataset on http://archive.ics.uci.edu/ml/datasets/Tic-Tac-Toe+Endgame.
Part 2. Mushroom Classification
This part includes code for logistic regression model used to perform "classification of mushrooms" problem.
Survey and implementation of Differentially private M-estimator
This project includes implementation and analysis of differentially private M-estimator. Based on paper:- https://papers.nips.cc/paper/4376-differentially-private-m-estimators.pdf