Privacy-preserving describes methods and systems designed to perform useful computation or data analysis while limiting exposure of the underlying personal or sensitive data.
Semantic Classification
Content
- Privacy-preserving techniques include differential privacy, which bounds how much a single record can affect an output, homomorphic encryption, which computes on encrypted data, and secure multi-party computation, which lets parties jointly compute a function without revealing their inputs.
- These methods are applied in machine learning, analytics, and data sharing to meet confidentiality requirements and regulatory constraints while still extracting value from data.