Privacy-Preserving Machine Learning (SpringerBriefs on Cyber Security Systems and Networks)

✍ Scribed by Jin Li, Ping Li, Zheli Liu, Xiaofeng Chen, Tong Li

• Publisher: Springer
• Year: 2022
• Tongue: English
• Leaves: 95
• Category: Library

✦ Synopsis

This book provides a thorough overview of the evolution of privacy-preserving machine learning schemes over the last ten years, after discussing the importance of privacy-preserving techniques. In response to the diversity of Internet services, data services based on machine learning are now available for various applications, including risk assessment and image recognition. In light of open access to datasets and not fully trusted environments, machine learning-based applications face enormous security and privacy risks. In turn, it presents studies conducted to address privacy issues and a series of proposed solutions for ensuring privacy protection in machine learning tasks involving multiple parties. In closing, the book reviews state-of-the-art privacy-preserving techniques and examines the security threats they face.

✦ Table of Contents

Preface
Contents
1 Introduction
1.1 What Is Machine Learning?
1.2 Why Machine Learning Needs Privacy-Preserving Manner
1.3 Security Threats
1.4 Bibliographic Notes
References
2 Secure Cooperative Learning in Early Years
2.1 An Overview of Neural Network
2.2 Back-Propagation Learning
2.3 Vertically Partitioned Training Dataset
2.3.1 Privacy-Preserving Two-Party Training
2.3.2 Secure Manner
2.3.3 Scheme Details
2.3.4 Analysis of Security and Accuracy Loss
2.4 Arbitrarily Partitioned Training Dataset
2.4.1 BGN Homomorphic Encryption
2.4.2 Overviews
2.4.3 Scheme Details
References
3 Outsourced Computation for Learning
3.1 Outsourced Computation
3.2 Multi-key Privacy-Preserving Deep Learning
3.2.1 Deep Learning
3.2.2 Homomorphic Encryption with Double Decryption Mechanism
3.2.3 Basic Scheme
3.2.4 Advance Scheme
3.2.5 Security Analysis
References
4 Secure Distributed Learning
4.1 Distributed Privacy-Preserving Deep Learning
4.1.1 Distributed Selective SGD
4.1.2 Scheme Details
4.2 Secure Aggregation for Deep Learning
4.2.1 Secure Manner
4.2.2 Technical Intuition
4.2.3 Secure Protocol
References
5 Learning with Differential Privacy
5.1 Differential Privacy
5.1.1 Definition
5.1.2 Privacy Mechanism
5.2 Deep Learning with Differential Privacy
5.2.1 Differentially Private SGD Algorithm
5.2.2 Privacy Account
5.3 Distributed Deep Learning with Differential Privacy
5.3.1 Private Algorithm
5.3.2 Estimating Sensitivity
References
6 Applications—Privacy-Preserving Image Processing
6.1 Machine Learning Image Processing for Privacy Protection
6.2 Feature Extraction Methods of Machine Learning Image Processing
6.3 Main Models of Machine Learning Image Processing for Privacy Protection
6.3.1 Privacy-Preserving Face Recognition
6.3.2 Privacy-Preserving Object Recognition
6.3.3 Privacy-Preserving Classification
Reference
7 Threats in Open Environment
7.1 Data Reconstruction Attack
7.1.1 Threat Model
7.1.2 Attack Method
7.2 Membership Inference Attack
7.2.1 Threat Model
7.2.2 Attack Method
7.3 Model Stealing Attack
7.3.1 Threat Model
7.3.2 Attack Method
References
8 Conclusion