Note: Intro -- inside front cover -- Privacy-Preserving Machine Learning -- Copyright -- contents -- front matter -- preface -- acknowledgments -- about this book -- Who should read this book -- How this book is organized: A road map -- About the code -- liveBook discussion forum -- about the authors -- about the cover illustration -- Part 1 Basics of privacy-preserving machine learning with differential privacy -- 1 Privacy considerations in machine learning -- 1.1 Privacy complications in the AI era -- 1.2 The threat of learning beyond the intended purpose -- 1.2.1 Use of private data on the fly -- 1.2.2 How data is processed inside ML algorithms -- 1.2.3 Why privacy protection in ML is important -- 1.2.4 Regulatory requirements and the utility vs. privacy tradeoff -- 1.3 Threats and attacks for ML systems -- 1.3.1 The problem of private data in the clear -- 1.3.2 Reconstruction attacks -- 1.3.3 Model inversion attacks -- 1.3.4 Membership inference attacks -- 1.3.5 De-anonymization or re-identification attacks -- 1.3.6 Challenges of privacy protection in big data analytics -- 1.4 Securing privacy while learning from data: Privacy-preserving machine learning -- 1.4.1 Use of differential privacy -- 1.4.2 Local differential privacy -- 1.4.3 Privacy-preserving synthetic data generation -- 1.4.4 Privacy-preserving data mining techniques -- 1.4.5 Compressive privacy -- 1.5 How is this book structured? -- Summary -- 2 Differential privacy for machine learning -- 2.1 What is differential privacy? -- 2.1.1 The concept of differential privacy -- 2.1.2 How differential privacy works -- 2.2 Mechanisms of differential privacy -- 2.2.1 Binary mechanism (randomized response) -- 2.2.2 Laplace mechanism -- 2.2.3 Exponential mechanism -- 2.3 Properties of differential privacy -- 2.3.1 Postprocessing property of differential privacy. , 2.3.2 Group privacy property of differential privacy -- 2.3.3 Composition properties of differential privacy -- Summary -- 3 Advanced concepts of differential privacy for machine learning -- 3.1 Applying differential privacy in machine learning -- 3.1.1 Input perturbation -- 3.1.2 Algorithm perturbation -- 3.1.3 Output perturbation -- 3.1.4 Objective perturbation -- 3.2 Differentially private supervised learning algorithms -- 3.2.1 Differentially private naive Bayes classification -- 3.2.2 Differentially private logistic regression -- 3.2.3 Differentially private linear regression -- 3.3 Differentially private unsupervised learning algorithms -- 3.3.1 Differentially private k-means clustering -- 3.4 Case study: Differentially private principal component analysis -- 3.4.1 The privacy of PCA over horizontally partitioned data -- 3.4.2 Designing differentially private PCA over horizontally partitioned data -- 3.4.3 Experimentally evaluating the performance of the protocol -- Summary -- Part 2 Local differential privacy and synthetic data generation -- 4 Local differential privacy for machine learning -- 4.1 What is local differential privacy? -- 4.1.1 The concept of local differential privacy -- 4.1.2 Randomized response for local differential privacy -- 4.2 The mechanisms of local differential privacy -- 4.2.1 Direct encoding -- 4.2.2 Histogram encoding -- 4.2.3 Unary encoding -- Summary -- 5 Advanced LDP mechanisms for machine learning -- 5.1 A quick recap of local differential privacy -- 5.2 Advanced LDP mechanisms -- 5.2.1 The Laplace mechanism for LDP -- 5.2.2 Duchi's mechanism for LDP -- 5.2.3 The Piecewise mechanism for LDP -- 5.3 A case study implementing LDP naive Bayes classification -- 5.3.1 Using naive Bayes with ML classification -- 5.3.2 Using LDP naive Bayes with discrete features -- 5.3.3 Using LDP naive Bayes with continuous features. , 5.3.4 Evaluating the performance of different LDP protocols -- Summary -- 6 Privacy-preserving synthetic data generation -- 6.1 Overview of synthetic data generation -- 6.1.1 What is synthetic data? Why is it important? -- 6.1.2 Application aspects of using synthetic data for privacy preservation -- 6.1.3 Generating synthetic data -- 6.2 Assuring privacy via data anonymization -- 6.2.1 Private information sharing vs. privacy concerns -- 6.2.2 Using k-anonymity against re-identification attacks -- 6.2.3 Anonymization beyond k-anonymity -- 6.3 DP for privacy-preserving synthetic data generation -- 6.3.1 DP synthetic histogram representation generation -- 6.3.2 DP synthetic tabular data generation -- 6.3.3 DP synthetic multi-marginal data generation -- 6.4 Case study on private synthetic data release via feature-level micro-aggregation -- 6.4.1 Using hierarchical clustering and micro-aggregation -- 6.4.2 Generating synthetic data -- 6.4.3 Evaluating the performance of the generated synthetic data -- Summary -- Part 3 Building privacy-assured machine learning applications -- 7 Privacy-preserving data mining techniques -- 7.1 The importance of privacy preservation in data mining and management -- 7.2 Privacy protection in data processing and mining -- 7.2.1 What is data mining and how is it used? -- 7.2.2 Consequences of privacy regulatory requirements -- 7.3 Protecting privacy by modifying the input -- 7.3.1 Applications and limitations -- 7.4 Protecting privacy when publishing data -- 7.4.1 Implementing data sanitization operations in Python -- 7.4.2 k-anonymity -- 7.4.3 Implementing k-anonymity in Python -- Summary -- 8 Privacy-preserving data management and operations -- 8.1 A quick recap of privacy protection in data processing and mining -- 8.2 Privacy protection beyond k-anonymity -- 8.2.1 l-diversity -- 8.2.2 t-closeness. , 8.2.3 Implementing privacy models with Python -- 8.3 Protecting privacy by modifying the data mining output -- 8.3.1 Association rule hiding -- 8.3.2 Reducing the accuracy of data mining operations -- 8.3.3 Inference control in statistical databases -- 8.4 Privacy protection in data management systems -- 8.4.1 Database security and privacy: Threats and vulnerabilities -- 8.4.2 How likely is a modern database system to leak private information? -- 8.4.3 Attacks on database systems -- 8.4.4 Privacy-preserving techniques in statistical database systems -- 8.4.5 What to consider when designing a customizable privacy-preserving database system -- Summary -- 9 Compressive privacy for machine learning -- 9.1 Introducing compressive privacy -- 9.2 The mechanisms of compressive privacy -- 9.2.1 Principal component analysis (PCA) -- 9.2.2 Other dimensionality reduction methods -- 9.3 Using compressive privacy for ML applications -- 9.3.1 Implementing compressive privacy -- 9.3.2 The accuracy of the utility task -- 9.3.3 The effect of ρ' in DCA for privacy and utility -- 9.4 Case study: Privacy-preserving PCA and DCA on horizontally partitioned data -- 9.4.1 Achieving privacy preservation on horizontally partitioned data -- 9.4.2 Recapping dimensionality reduction approaches -- 9.4.3 Using additive homomorphic encryption -- 9.4.4 Overview of the proposed approach -- 9.4.5 How privacy-preserving computation works -- 9.4.6 Evaluating the efficiency and accuracy of the privacy-preserving PCA and DCA -- Summary -- 10 Putting it all together: Designing a privacy-enhanced platform (DataHub) -- 10.1 The significance of a research data protection and sharing platform -- 10.1.1 The motivation behind the DataHub platform -- 10.1.2 DataHub's important features -- 10.2 Understanding the research collaboration workspace -- 10.2.1 The architectural design. , 10.2.2 Blending different trust models -- 10.2.3 Configuring access control mechanisms -- 10.3 Integrating privacy and security technologies into DataHub -- 10.3.1 Data storage with a cloud-based secure NoSQL database -- 10.3.2 Privacy-preserving data collection with local differential privacy -- 10.3.3 Privacy-preserving machine learning -- 10.3.4 Privacy-preserving query processing -- 10.3.5 Using synthetic data generation in the DataHub platform -- Summary -- Appendix A. More details about differential privacy -- A.1 The formal definition of differential privacy -- A.2 Other differential privacy mechanisms -- A.2.1 Geometric mechanism -- A.2.2 Gaussian mechanism -- A.2.3 Staircase mechanism -- A.2.4 Vector mechanism -- A.2.5 Wishart mechanism -- A.3 Formal definitions of composition properties of DP -- A.3.1 The formal definition of sequential composition DP -- A.3.2 The formal definition of parallel composition DP -- references -- Appendix -- index -- inside back cover.