14 June 2019 – Associate Professor Xiao Xiaokui and PhD students Bao Ergute and Zhao Xuejun won third place in the final round of the 2018 Differential Privacy Synthetic Data Challenge. The challenge lasted from 31 October 2018 to 23 April 2019 and the results were announced on 20 May 2019.
The Differential Privacy Synthetic Challenge, held from October 2018 to April 2019, was organised by the United States of America’s National Institute of Standards and Technology. Over 87 researchers from around the world participated in the three marathon matches. Teams were tasked to create new or improved differentially private data generation tools.
Privacy preserving tools are used to create artificial data to protect the privacy of the original sensitive data. These tools remove any sensitive personal information found in the original data sets by producing data that is similar to its original source but does not contain traces of information linked to specific people. Teams undertaking this challenge had to design new algorithms for generating synthetic data and prove that the synthetic data produced satisfies differential privacy. The developed algorithms were then compared against each other over their empirical accuracy to the real data.
Prof Xiao, Ergute and Xuejun developed PrivBayes, a differentially private method of releasing high dimensional data. The team partnered with Alibaba research scientist Dr Ding Bolin and Nanyang Technological University Assistant Professor Zhao Jun for the competition.
“Throughout the competition, the team and I were very much busy with our research, teaching, and school commitments. It was challenging to squeeze time out of our schedules to design and implement our PrivBayes algorithm. This resulted in a number of sleepless nights for us. However, we felt that our efforts have truly paid off and it was truly exciting to have won third place,” said Prof Xiao.
Associate Professor Xiao Xiaokui joined the Department of Computer Science in NUS Computing on 25 January 2018. His research interests are in the areas of data management, especially data privacy and algorithms for large data. Ergute and Xuejun are currently PhD students pursuing research in Database under the supervision of Prof Xiao.