Federated learning with Python : design and implement a federated learning system and develop applications using existing frameworks
Learn the essential skills for building an authentic federated learning system with Python and take your machine learning applications to the next level Key Features Design distributed systems that can be applied to real-world federated learning applications at scale Discover multiple aggregation sc...
Enregistré dans:
| Auteur principal: | |
|---|---|
| Format: | Livre papier |
| Langue: | Anglais |
| Publié: |
Birmingham :
Packt Publishing Limited
2022.
|
| Sujets: | |
| Autres localisations: | Voir dans le Sudoc |
| Edition sous un autre format: | • Federated Learning with Python, Design and implement a federated learning system and develop applications using existing frameworks, Kiyoshi Nakayama, George Jeno, Birmingham, Packt Publishing, 2022, 978-18-0324-875-2 |
Table des matières:
- Cover Title Page Copyright and Credits Acknowledgments Contributors Table of Contents Preface Part 1 Federated Learning Conceptual Foundations Chapter 1: Challenges in Big Data and Traditional AI Understanding the nature of big data Definition of big data Big data now Triple-A mindset for big data Data privacy as a bottleneck Risks in handling private data Increased data protection regulations From privacy by design to data minimalism Impacts of training data and model bias Expensive training of big data Model bias and training data Model drift and performance degradation How models can stop working Continuous monitoring the price of letting causation go FL as the main solution for data problems Summary Further reading Chapter 2: What Is Federated Learning? Understanding the current state of ML What is a model? ML automating the model creation process Deep learning Distributed learning nature toward scalable AI Distributed computing Distributed ML Edge inference Edge training Understanding FL Defining FL The FL process FL system considerations Security for FL systems Decentralized FL and blockchain Summary Further reading Chapter 3: Workings of the Federated Learning System FL system architecture Cluster aggregators Distributed agents Database servers Intermediate servers for low computational agent devices Understanding the FL system flow from initialization to continuous operation Initialization of the database, aggregator, and agent Initial model upload process by initial agent Overall FL cycle and process of the FL system Synchronous and asynchronous FL The aggregator-side FL cycle and process The agent-side local retraining cycle and process Model interpretation based on deviation from baseline outputs Basics of model aggregation What exactly does it mean to aggregate models? FedAvg Federated averaging Furthering scalability with horizontal design Horizontal design with semi-global model Distributed database Asynchronous agent participation in a multiple-aggregator scenario Semi-global model synthesis Summary Further reading Part 2 The Design and Implementation of the Federated Learning System Chapter 4: Federated Learning Server Implementation with Python Technical requirements Main software components of the aggregator and database Aggregator-side codes lib/util codes Database-side code Toward the configuration of the aggregator Implementing FL server-side functionalities Importing libraries for the FL server Defining the FL Server class Initializing the FL server Registration function of agents The server for handling messages from local agents The global model synthesis routine Functions to send the global models to the agents