Demonstration of Damson: Differential Privacy for Analysis of Large Data

Abstract

We demonstrate Damson, a novel and powerful tool for publishing the results of biomedical research with strong privacy guarantees. Damson is developed based on the theory of differential privacy, which ensures that the adversary cannot infer the presence or absence of any individual from the published results, even with substantial background knowledge. Damson supports a variety of analysis tasks that are common in biomedical studies, including histograms, marginals, data cubes, classification, regression, clustering, and ad-hoc selection-counts. Additionally, Damson contains an effective query optimization engine, which obtains high accuracy for analysis results, while minimizing the privacy costs of performing such analysis.

DOI: 10.1109/ICPADS.2012.137

Extracted Key Phrases

4 Figures and Tables

Cite this paper

@article{Winslett2012DemonstrationOD, title={Demonstration of Damson: Differential Privacy for Analysis of Large Data}, author={Marianne Winslett and Yin Yang and Zhenjie Zhang}, journal={2012 IEEE 18th International Conference on Parallel and Distributed Systems}, year={2012}, pages={840-844} }