Privately Answer Count-Queries on Periodical Datasets
Gang Li
senior lecturer
Deakin University
Abstract: Most privacy-preserving studies focus on static dataset releases or queries. However, in practice, people usually release these datasets periodically. Differential privacy provides a possible way to deal with continual releases by simplifying the continual dataset release to a bit stream release. This method does not consider the coupling between events and records. In fact, the coupled information might reveal more private information than expected. This work proposes a privacy-preserving mechanism that explicitly deals with the coupled information in the periodically generated datasets. It decomposes the coupled information into intra-coupled information and inter-coupled information, and proposes the notion of coupled sensitivity. A mechanism, CCR, is developed for privately answering a large set of count query. The initial performance outperforms the naive differential privacy mechanism when answering a large set of queries.
Bio: Gang Li, visiting scholar to NJU, senior lecturer, director of computer science course, and HDR coordinator for Faculty of Science and Technology in Deakin University. His recent research interests are in the area of data privacy, and private data release.