Differential Privacy is a privacy framework that provides a mathematical guarantee that data has been completely deidentified. What does that mean? How do you do it? And how do you do it well or poorly?
Synthetic data imitates real population data using synthetic people. How does that work? How can you tell if synthetic data is safe to use in place of the original data? This presentation answers such questions. It introduces the CenSyn data synthesizer and evaluator that we are developing for the American Community Survey for the US Census Bureau. From the 2020 Confidentiality and Data Access Webinar Series, jointly organized by the Council of Professional Associations on Federal Statistics (COPFAS) and the Federal Committee on Statistical Methodology (FCSM).
We protect individual privacy using innovative techniques, including statistical modeling and privacy-preserving machine learning. Our synthetic data generators create fully de-identified data that retains the key characteristics of your original data. This enables you to share the synthetic, privatized data with third-parties safely and securely, and conduct analytics to effectively reach the same conclusions that you would with your original data. We developed and field tested these tools at the US Census Bureau, as part of disclosure modernization efforts for the American Community Survey and American Housing Survey. The same proven tools can efficiently and effectively meet your data privacy needs. We can help you maximize the business value from your data by standardizing and streamlining synthetic data generation, and enabling risk-free data sharing.
KRC researches and engineers robust analytics tools that are designed to maintain accuracy over privatized, noisy data. As synthetic and privatized data see more widespread adoption, beginning with the use of Differential Privacy in the 2020 US Decennial Census, being able to work effectively with this data will become business critical. Leveraging analytics that are robust against privatization noise can help you overcome unanticipated data challenges that may arise. We have the know-how and the track record to help you navigate the emerging privacy-regulatory landscape.
KRC has developed rigorous scientific methods for evaluating the performance of privacy-preserving algorithms, software, and other technologies. As technical leads on the National Institute of Standards and Technology (NIST) Differential Privacy Synthetic Data Challenge and Differential Privacy Temporal Map Challenge, we help the NIST design challenge problems, test data, evaluation metrics, and fair comparative evaluation strategies on real world practical applications. We’ve untangled complexities in both algorithm development and evaluation, explaining the mathematical nuances of differential privacy for challenge organizers and participants, working with stakeholders from the data user community to ensure that utility measures are faithful to data users’ needs. KRC can do the same to help your organization navigate new privacy technologies.
KRC researches and develops innovative privacy-preserving technologies to address an array of challenges. For instance, a mobile phone application we developed for the DARPA Brandeis program, provides early warning to first responders at the onset of a crisis. The noise-resistant metrics that we developed for this system operate over privatized cell phone data to accurately predict the onset of a crisis, enabling a more timely potentially life-saving response. This is just one example of a successful application of our research, and we can do the same for you. Building on our proven techniques, we can develop and demonstrate such privacy applications for your data.