Federated learning

• Thesis title in Czech: Kolaborativní učení
• Thesis title in English: Federated learning
• Key words: federativní učení|strojové učení|kolaborativní učení
• English key words: federated learning|machine learning|collaborative learning
• Academic year of topic announcement: 2022/2023
• Thesis type: Bachelor's thesis
• Thesis language: angličtina
• Department: Department of Logic (21-KLOG)
• Supervisor: Mgr. Petr Švarný, Ph.D.
• Author: hidden - assigned and confirmed by the Study Dept.
• Date of registration: 10.11.2022
• Date of assignment: 10.11.2022
• Administrator's approval: approved
• Confirmed by Study dept. on: 17.12.2022
• Date and time of defence: 16.06.2023 09:00
• Date of electronic submission: 09.05.2023
• Date of proceeded defence: 16.06.2023
• Submitted/finalized: committed by student and finalized
• Opponents: Mgr. Martin Blicha, Ph.D.

Guidelines

A common problem in Machine Learning is the necessity of large datasets for the training of the algorithms.
While there are many approaches to solve this problem, one possible solution is the distribution of the training datasets and then the joining of separately trained models.

The student must research and summarize the state of the art of federated training, present its general idea and related notions, and discuss the main known solutions.This analysis should discuss the basic theoretical framework of federated learning.
It should focus on the effect of the number of data sources used for training (e.g., differences between using three, ten or even more data sources, what is the performance effect and what are the arising challenges).

While a thorough analysis of the options would be the main goal of the thesis, the thesis should also attempt to implement a solution of federated learning.
This implemented solution should either present a novel approach that would be motivated in the theoretical part of the thesis or implement one of the already known and analyzed approaches.
In the latter case, the motivation for the choice must be well-explained in the text.It can be a prototype if the student fully implements the solution.

The thesis could additionally or alternatively present a comparison of multiple techniques of distributed training.

If appropriate, the thesis can also discuss the practical use of such solutions, e.g., the need to use encryption for sensitive data.

Any analysis, the initial theoretical analysis or performance analysis of the implemented solutions, must be accompanied by clear and appropriate visualizations.

References

Kairouz, P., McMahan, H. B., Avent, B., Bellet, A., Bennis, M., Bhagoji, A. N., ... & Zhao, S. (2021). Advances and open problems in federated learning. Foundations and Trends® in Machine Learning, 14(1–2), 1-210.
Li, L., Fan, Y., Tse, M., & Lin, K. Y. (2020). A review of applications in federated learning. Computers & Industrial Engineering, 149, 106854.
Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep learning. MIT press.
Deppeler, A. (2020). Automated Machine Learning and Federated Learning. The AI Book: The Artificial Intelligence Handbook for Investors, Entrepreneurs and FinTech Visionaries, 248-250.
Yang, Q., Fan, L., & Yu, H. (Eds.). (2020). Federated Learning: Privacy and Incentive (Vol. 12500). Springer Nature.