Privacy-preserving data

The paper presents a new approach, PrivPGD, for differentially private data synthesis of protected tabular datasets, an important task in sensitive domains like healthcare and government. Existing methods typically use…