Learning from Functionality Outputs: Private Join and Compute in the Real World

Artifact description (availability)

In this artifact, we include the code for all four of our attacks (the search tree, MLE, DFT, and compressed sensing attacks) and instructions for reproducing our results. Our artifact is structured as follows. In the top-level directory, there are two directories: (1) “search_tree” which contains the implementation, benchmarking script, and random data generation for the search-tree attack, and (2) “cs_dft_mle” which contains the implementation, benchmarking script, and random data generation for the compressed sensing, discrete fourier transform, and maximum-likelihood estimation attacks. Each of these two directories contains a file named “requirements.txt” which contains a list of all the dependencies needed for the respective attacks and a “README.md” file that provides a detailed description of how to run the code and reproduce our experiments.

Funding

Francesca Falzon is supported by Armasuisse Science and Technology.

Tianxin Tang is supported by an NWO VIDI grant (Project No. VI.Vidi.193.066).