Privacy-preserving Technologies for Artificial Intelligence

作者：Murat Kantarcioglu · 发表于：River Publishers eBooks · 年份：2026 · DOI：10.1201/9788743812074-14 · 研究领域：Privacy-Preserving Technologies in Data、Big Data and Digital Economy、Adversarial Robustness in Machine Learning

Privacy remains a significant barrier to unlocking the full potential of AI in sensitive domains such as healthcare. In this chapter, we survey methods for preserving privacy while enabling useful data analysis. We review anonymization approaches, including k-anonymity, differential privacy, and federated learning, and explain the strengths and weaknesses of each. The chapter emphasizes the trade-offs between utility, privacy, and computational cost, and illustrates how these trade-offs manifest in practical deployment. We also discuss real-world examples that demonstrate how privacy-preserving AI can facilitate innovation without compromising individual protections.