Administrative Information
| Title | Theory of Federated Learning (Profiling and Personalization) |
| Duration | 45-60 min |
| Module | C |
| Lesson Type | Lecture |
| Focus | Technical - Future AI |
| Topic | Advances in ML models through a HC lens - A result Oriented Study |
Keywords
Federated Learning, knowledge-based system, privacy preservation,
Learning Goals
- Provide the motivations for doing Federated Learning
- Provide an initial understanding of the basic techniques for Federated Learning
- Discuss the main limitations and the challenges connected to them
Expected Preparation
Learning Events to be Completed Before
Obligatory for Students
- Basic knowledge of probability and statistics
- Basic understanding of deep learning training (SGD, backpropagation algorithm) and evaluation techniques
- Introduction to machine learning and deep learning concepts given in previous lectures
Optional for Students
None.
References and background for students
None.
Recommended for Teachers
- P. Kairouz, et al. "Advances and Open Problems in Federated Learning", Foundations and Trends® in Machine Learning: Vol. 14: No. 1–2, pp 1-210. - URL
- Konečný, Jakub and McMahan, H. Brendan and Ramage, Daniel and Richtárik, Peter. "Federated Optimization: Distributed Machine Learning for On-Device Intelligence". arXiv 2016.
- Chen Zhang, et al., A survey on federated learning, Knowledge-Based Systems, Volume 216, 2021, 106775, ISSN 0950-7051 - URL
- Google Federated Learning and AI
- Federated Learning: Collaborative Machine Learning without Centralized Training Data
- Deep learning -> Federated learning in 10 lines
- tensorflow/federated
Lesson materials
Instructions for Teachers
- Provide an overview of the techniques, their pros and cons
- Propose pop-up quizzes
The learning event shall refer to model types, their evaluation and possible optimization techniques.
Outline
| Duration | Description | Concepts |
|---|---|---|
| 10 min | Introduction: Motivating scenario and introduction to federated learning: what it is, what it is for, when and why it is needed. | Data gravity, data privacy and the definition of enabling scenario. |
| 10 min | Federated Learning: basic concepts, system definition and algorithmic overview | Basic notions of the Federated Learning approach |
| 15 min | Federated Average algorithm: Formal definition and properties | Basic algorithm for federated learning |
| 20 min | Beyond federated average: limitations of federated average, challenges and possible solutions. | Data imbalance, personalisation, fairness |
| 5 min | Conclusion, questions and answers | Summary |
Acknowledgements
The Human-Centered AI Masters programme was Co-Financed by the Connecting Europe Facility of the European Union Under Grant №CEF-TC-2020-1 Digital Skills 2020-EU-IA-0068.