**********   Requirements:   **********

- The code works with Python3.
- Datasets: All of the datasets used are given in the datasets folder except for the cifar-10 dataset. Cifar-10 dataset can be downloaded from the link "http://www.cs.toronto.edu/~kriz/cifar.html" by clicking on "CIFAR-10 python version". The downloaded zip file must extracted into the "datasets" folder, and the provided code will take care of the rest. Cifar-10 is used only for Figure 2, the datasets for the remaining figures are already uploaded in the supplementary material.
- The required python packages are as follows (all of the packages be installed with pip3 install package-name): numpy, matplotlib, sklearn, dppy.



**********   Commands to generate the figures in the paper:   **********

- For figure 1, run "python3 dist_sketching_dpp.py 1 boston 20 none"
- For figure 2a, run "python3 dist_sketching_dpp.py 2 cifar 1000 gaus"
  For figure 2b, run "python3 dist_sketching_dpp.py 2 cifar 1000 unif"
  For figure 2c, run "python3 dist_sketching_dpp.py 2 cifar 1000 surrogate_p1"
- For figure 3a, run "python3 dist_sketching_dpp.py 3 statlog-australian-credit 50 none"
  For figure 3b, run "python3 dist_sketching_dpp.py 3 breast-cancer-wisc 50 none"
  For figure 3c, run "python3 dist_sketching_dpp.py 3 ionosphere 100 none"
- For figure 4, run "python3 dist_sketching_dpp.py 4 boston 50 none"
- For figure 5, run "python3 dist_sketching_dpp.py 5 statlog-australian-credit 50 surrogate_p1"

- To generate figures with any other desired parameters, the command works as follows: "python3 dist_sketching_dpp.py figure_no dataset_name sketch_size sketch_type"