Muutke küpsiste eelistusi

Federated Learning [Pehme köide]

4.20/5 (8 hinnangut Goodreads-ist)
, , , , ,
Teised raamatud teemal:
Federated LearningSuurem pilt
Teised raamatud teemal:
Püsilink: https://www.kriso.ee/db/9781681736976.html
This book shows how federated machine learning allows multiple data owners to collaboratively train and use a shared prediction model while keeping all the local training data private. Traditional machine learning approaches need to combine all data at one location, typically a data center, which may very well violate the laws on user privacy and data confidentiality. Today, many parts of the world demand that technology companies treat user data carefully according to user-privacy laws. The European Union's General Data Protection Regulation (GDPR) is a prime example.

In this book, we describe how federated machine learning addresses this problem with novel solutions combining distributed machine learning, cryptography and security, and incentive mechanism design based on economic principles and game theory. We explain different types of privacy-preserving machine learning solutions and their technological backgrounds, and highlight some representative practical use cases. We show how federated learning can become the foundation of next-generation machine learning that caters to technological and societal needs for responsible AI development and application.
Preface xiii
Acknowledgments xvii
1 Introduction
1(16)
1.1 Motivation
1(2)
1.2 Federated Learning as a Solution
3(7)
1.2.1 The Definition of Federated Learning
4(3)
1.2.2 Categories of Federated Learning
7(3)
1.3 Current Development in Federated Learning
10(5)
1.3.1 Research Issues in Federated Learning
10(1)
1.3.2 Open-Source Projects
11(2)
1.3.3 Standardization Efforts
13(1)
1.3.4 The Federated AI Ecosystem
14(1)
1.4 Organization of this Book
15(2)
2 Background
17(16)
2.1 Privacy-Preserving Machine Learning
17(1)
2.2 PPML and Secure Ml
17(1)
2.3 Threat and Security Models
18(3)
2.3.1 Privacy Threat Models
18(3)
2.3.2 Adversary and Security Models
21(1)
2.4 Privacy Preservation Techniques
21(12)
2.4.1 Secure Multi-Party Computation
21(5)
2.4.2 Homomorphic Encryption
26(3)
2.4.3 Differential Privacy
29(4)
3 Distributed Machine Learning
33(16)
3.1 Introduction to DML
33(3)
3.1.1 The Definition of DML
33(2)
3.1.2 DML Platforms
35(1)
3.2 Scalability-Motivated DML
36(4)
3.2.1 Large-Scale Machine Learning
36(1)
3.2.2 Scalability-Oriented DML Schemes
37(3)
3.3 Privacy-Motivated DML
40(5)
3.3.1 Privacy-Preserving Decision Trees
40(2)
3.3.2 Privacy-Preserving Techniques
42(1)
3.3.3 Privacy-Preserving DML Schemes
42(3)
3.4 Privacy-Preserving Gradient Descent
45(3)
3.4.1 Vanilla Federated Learning
45(1)
3.4.2 Privacy-Preserving Methods
46(2)
3.5 Summary
48(1)
4 Horizontal Federated Learning
49(20)
4.1 The Definition of HFL
49(1)
4.2 Architecture of HFL
50(5)
4.2.1 The Client-Server Architecture
51(2)
4.2.2 The Peer-to-Peer Architecture
53(1)
4.2.3 Global Model Evaluation
54(1)
4.3 The Federated Averaging Algorithm
55(7)
4.3.1 Federated Optimization
55(3)
4.3.2 The FedAvg Algorithm
58(2)
4.3.3 The Secured FedAvg Algorithm
60(2)
4.4 Improvement of the FedAvg Algorithm
62(2)
4.4.1 Communication Efficiency
62(2)
4.4.2 Client Selection
64(1)
4.5 Related Works
64(2)
4.6 Challenges and Outlook
66(3)
5 Vertical Federated Learning
69(14)
5.1 The Definition of VFL
69(2)
5.2 Architecture of VFL
71(2)
5.3 Algorithms of VFL
73(8)
5.3.1 Secure Federated Linear Regression
73(3)
5.3.2 Secure Federated Tree-Boosting
76(5)
5.4 Challenges and Outlook
81(2)
6 Federated Transfer Learning
83(12)
6.1 Heterogeneous Federated Learning
83(1)
6.2 Federated Transfer Learning
84(2)
6.3 Ihe FTL Framework
86(6)
6.3.1 Additively Homomorphic Encryption
88(1)
6.3.2 The FTL Training Process
89(1)
6.3.3 The FTL Prediction Process
90(1)
6.3.4 Security Analysis
90(1)
6.3.5 Secret Sharing-Based FTL
91(1)
6.4 Challenges and Outlook
92(3)
7 Incentive Mechanism Design for Federated Learning
95(12)
7.1 Paying for Contributions
95(3)
7.1.1 Profit-Sharing Games
95(2)
7.1.2 Reverse Auctions
97(1)
7.2 A Fairness-Aware Profit Sharing Framework
98(5)
7.2.1 Modeling Contribution
98(1)
7.2.2 Modeling Cost
99(1)
7.2.3 Modeling Regret
100(1)
7.2.4 Modeling Temporal Regret
100(1)
7.2.5 The Policy Orchestrator
100(3)
7.2.6 Computing Payoff Weightage
103(1)
7.3 Discussions
103(4)
8 Federated Learning for Vision, Language, and Recommendation
107(14)
8.1 Federated Learning for Computer Vision
107(4)
8.1.1 Federated CV
107(2)
8.1.2 Related Works
109(1)
8.1.3 Challenges and Oudook
110(1)
8.2 Federated Learning for NLP
111(3)
8.2.1 Federated NLP
112(1)
8.2.2 Related Works
113(1)
8.2.3 Challenges and Oudook
114(1)
8.3 Federated Learning for Recommendation Systems
114(7)
8.3.1 Recommendation Model
115(1)
8.3.2 Federated Recommendation System
116(2)
8.3.3 Related Works
118(1)
8.3.4 Challenges and Oudook
118(3)
9 Federated Reinforcement Learning
121(12)
9.1 Introduction to Reinforcement Learning
121(3)
9.1.1 Policy
122(1)
9.1.2 Reward
122(1)
9.1.3 Value Function
122(1)
9.1.4 Model of the Environment
123(1)
9.1.5 RL Example
123(1)
9.2 Reinforcement Learning Algorithms
124(1)
9.3 Distributed Reinforcement Learning
124(2)
9.3.1 Asynchronous Distributed Reinforcement Learning
125(1)
9.3.2 Synchronous Distributed Reinforcement Learning
126(1)
9.4 Federated Reinforcement Learning
126(5)
9.5 Challenges and Outlook
131(2)
10 Selected Applications
133(10)
10.1 Finance
133(1)
10.2 Healthcare
134(2)
10.3 Education
136(1)
10.4 Urban Computing and Smart City
136(3)
10.5 Edge Computing and Internet of Things
139(1)
10.6 Blockchain
140(1)
10.7 5G MobUe Networks
141(2)
11 Summary and Outlook
143(2)
A Legal Development on Data Protection
145(10)
A.1 Data Protection in the European Union
145(6)
A.1.1 Ihe Terminology of GDPR
146(1)
A.1.2 Highlights of GDPR
147(3)
A.1.3 Impact of GDPR
150(1)
A.2 Data Protection in the USA
151(1)
A.3 Data Protection in China
152(3)
Bibliography 155(32)
Authors' Biographies 187