(Downloads - 0)
For more info about our services contact : help@bestpfe.com
Table of contents
Abstract
Resume [Francais]
Acknowledgements
Contents
List of Figures
List of Tables
Acronyms
Publications
1 Introduction
1.1 Machine Learning as a Service
1.2 Data Privacy vs Machine Learning Techniques
1.3 Privacy-preserving protocols for Machine Learning Techniques
1.4 Contributions
1.5 Organisation
2 Machine Learning Techniques
2.1 Neural Networks
2.1.1 Convolutional Layer
2.1.2 Fully Connected Layer
2.1.3 Activation Layer
2.1.4 Pooling Layer
2.1.5 Complementary Functions in Neural Networks
2.1.6 Neural Network Model Structure
2.1.7 Neural Networks Accuracy Evaluation
2.2 Clustering
2.2.1 k-means
2.2.2 DBSCAN
2.2.3 TRACLUS
2.2.4 Clustering Quality Evaluation
2.3 Data Aggregation
2.4 Summary of Machine Learning techniques
3 Cryptographic Techniques
3.1 Security Notions
3.1.1 Notations
3.1.2 Ideal/Real Simulation paradigm.
3.1.3 Adversarial Models and Attacks
3.1.4 Chosen Plaintext Attack
3.2 Secure Multi-party Computation
3.2.1 Oblivious Transfer
3.2.2 Yao’s Garbled Circuits
3.2.3 Secret Sharing
3.2.4 Available Libraries
3.3 Homomorphic Encryption
3.3.1 Partially Homomorphic Encryption
3.3.2 Somewhat Homomorphic Encryption
3.3.3 Fully Homomorphic Encryption
3.3.4 Available Libraries
3.4 Proxy Re-encryption
3.4.1 Homomorphic Proxy Re-encryption
3.5 Multi-key Fully Homomorphic Encryption
3.5.1 Asymmetric Multi-key Fully Homomorphic Encryption
3.5.2 Symmetric Multi-key Fully Homomorphic Encryption
3.6 Threshold Fully Homomorphic Encryption
3.7 Hybrid Protocol
3.8 Summary of Cryptographic techniques
I Privacy-preserving single-server machine learning techniques
4 Privacy-preserving Neural Network Classication
4.1 Introduction
4.2 Privacy vs. Neural Network
4.3 Prior Work
4.3.1 2PC-based solutions
4.3.2 HE-based solutions
4.3.3 Hybrid solutions
4.3.4 Solutions based on other cryptographic techniques
4.4 PAC: Privacy-preserving Arrhythmia Classication with neural networks .
4.4.1 Problem Statement
4.4.2 PAC: Description
4.4.3 Discussion on Principle Component Analysis
4.4.4 SIMD circuits
4.4.5 Implementation
4.4.6 Security Evaluation
4.4.7 Performance Evaluation
4.4.8 Summary
4.5 SwaNN: Switching among Cryptographic Tools for Privacy-Preserving Neural Network Predictions
4.5.1 Problem Statement
4.5.2 SwaNN: Description
4.5.3 Security Evaluation
4.5.4 Performance Evaluation
4.5.5 Summary
4.6 ProteiNN: Privacy-preserving one-to-many Neural Network classication .
4.6.1 Problem Statement
4.6.2 Threat Model
4.6.3 ProteiNN: Description
4.6.4 Security Evaluation
4.6.5 Performance Evaluation
4.6.6 Summary
4.7 Conclusion of privacy-preserving neural network classication
5 Privacy-preserving Clustering
5.1 Introduction
5.2 Privacy vs. Clustering
5.3 Prior Work
5.3.1 k-means
5.3.2 DBSCAN
5.3.3 Trajectory Analysis
5.4 pp-TRACLUS: Privacy-preserving TRAjectory CLUStering
5.4.1 pp-TRACLUS: Description
5.4.2 Security Evaluation
5.4.3 Performance Evaluation
5.4.4 Summary
5.5 Conclusion of privacy-preserving clustering
II Privacy-preserving two-server machine learning techniques
6 Privacy-preserving Neural Network Classication
6.1 Introduction
6.2 Prior Work
6.3 Two-server SwaNN
6.3.1 SwaNN: Description
6.3.2 Security Evaluation
6.3.3 Performance Evaluation
6.3.4 Summary
6.4 Conclusion of privacy-preserving neural network classication
7 Privacy-preserving Clustering
7.1 Introduction
7.2 Privacy vs. Clustering
7.3 Prior Work
7.4 Two-server pp-TRACLUS
7.4.1 Problem Statement
7.4.2 PHE-based pp-TRACLUS: Description
7.4.3 2PC-based pp-TRACLUS: Description
7.5 Conclusion of privacy-preserving clustering
8 Privacy-preserving Data Aggregation
8.1 Introduction
8.2 Privacy vs. Data Aggregation
8.3 Prior Work
8.3.1 DP (and HE)-based solutions
8.3.2 HE-based solutions
8.3.3 MPC/2PC-based solutions
8.3.4 Hybrid solutions
8.4 PRIDA: PRIvacy-preserving data aggregation with multiple Data Analysers
8.4.1 Problem Statement
8.4.2 PRIDA: Detailed description
8.4.3 Security Evaluation
8.4.4 Performance Evaluation
8.5 Conclusion of privacy-preserving data aggregation
9 Conclusion Remarks and Future Research
9.1 Summary
9.2 Future Work
Appendices
A Resume Francais
A.1 Apprentissage automatique en tant que service
A.2 Condentialite des donnees vs techniques d’apprentissage automatique .
A.3 Protocoles preservant la condentialite pour les techniques d’apprentissage automatique
A.4 Contributions
