High Performance Privacy Preserving AI

Artificial intelligence (AI) depends on data. In sensitive domains such as healthcare, security, finance, and many more there is therefore tension between unleashing the power of AI and maintaining the confidentiality and security of the relevant data. This book intended for researchers in academia and R&D engineers in industry explains how advances in three areas--AI, privacy-preserving techniques, and accelerationallow us to achieve the dream of high performance privacy-preserving AI. It also discusses applications enabled by this emerging interplay. The book covers techniques, specifically secure multi-party computation and homomorphic encryption, that provide complexity theoretic security guarantees even with a single data point. These techniques have traditionally been too slow for real-world usage, and the challenge is heightened with the large sizes of today's state-of-the-art neural networks, including large language models (LLMs). This book does not cover techniques like differential privacy that only concern statistical anonymization of data points.