DPSyn is a research group from Purdue University working on differential privacy. Their group has been conducting research on data privacy for about 15 years, with a focus on differential privacy for the most recent decade. The group has developed state-of-the-art algorithms for several tasks under the constraint of satisfying Differential Privacy and Local Differential Privacy. DPSyn is made up of Ninghui Li of West Lafayette, Indiana, USA, Zitao Li of China, and Tianhao Wang of China, all of whom are graduates of Purdue University.
DPSyn has a wealth of knowledge in differential privacy and have used this expertise to successfully participate in earlier prize competitions held by NIST. The team believes the 2020 Differential Privacy Temporal Map Challenge to be a good opportunity to think more about real world problems and further explore the designs of algorithms.
Their approach to this sprint was to modify the public dataset provided in the development phase with the privatized marginals from the target dataset. They first reweigh each record in the public dataset to limit the influence of each individual taxi_id. Then they split the privacy budget and query the target dataset to obtain the two-way marginal of pickup_community_area plus dropoff_community_area, one way marginal of company_id, and one way marginal of payment_type when privacy budget is small or two way marginal of pickup_community_area plus payment_type when the privacy budget is large. Then they replaced the attributes of records in the public datasets to match the privatized marginals, as depicted in the graphic below.
To contact this team, please email Ninghui Li at ninghui [at] cs.purdue.edu (ninghui[at]cs[dot]purdue[dot]edu).