Gaussian sketching

New research advances differential privacy in machine learning

Researchers have developed new methods to enhance differential privacy in machine learning, particularly for decentralized and causal structure learning. One approach, DPDL, uses a similarity-based calibration technique…