Previous Achievements of the Summer School

Over the years, the Federated Learning Summer School has successfully brought together leading researchers, industry experts, and students.

High-impact keynote talks on Federated Learning, Edge AI, and Data Privacy.
Hands-on workshops and coding sessions with state-of-the-art FL frameworks.
Collaborative research projects in security, healthcare, and IoT-based FL.
Networking opportunities for students, researchers, and industry leaders.

With each edition, the summer school continues to advance cutting-edge research and practical knowledge, equipping participants with the skills needed to contribute to the rapidly evolving field of Federated Learning.

Federated Learning Summer School Statistics

Participants Breakdown

Summer School Outcomes

Handling Non-IID Data in Federated Learning: An Experimental Evaluation Towards Unified Metrics

This study surveys strategies for handling Non-IID data in federated learning and introduces a new metric to assess data skew without accessing client data. The proposed metric aids in selecting effective strategies, enhancing both research and real-world applications.

MPCFL: Towards Multi-party Computation for Secure Federated Learning Aggregation

This study introduces MPCFL, a secure Federated Learning algorithm using multi-party computation and secret sharing to prevent data leakage. Evaluated on benchmarks, it enhances security and lays the foundation for robust privacy-preserving FL aggregation techniques.

Towards Accelerating the Adoption of Federated Learning for Heterogeneous Data

This study examines Federated Machine Learning (FML) for addressing data heterogeneity, privacy, and ownership challenges. It integrates the FEDMA algorithm and evaluates it on the FEMNIST dataset to simulate real-world heterogeneous AI scenarios.

Data Skew in Federated Learning: An Experimental Evaluation on Aggregation Algorithms

This study examines data skew challenges in Federated Learning, particularly for facial ethnicity classification with non-IID data. It evaluates FL aggregation algorithms and introduces an adaptive method to enhance model robustness, fairness, and privacy across diverse applications.

Bayesian Federated Learning with Stochastic Variational Inference

This study introduces Bayesian Federated Learning with Stochastic Variational Inference (BayFL-SVI) to improve non-IID data handling and model aggregation. It enhances accuracy and convergence rates, offering a strong foundation for future FL research and optimization.

FedPROM: A Zero-Trust Federated Learning Approach with Multi-Criteria Client Selection

This study introduces FedPROM, an MCDM-based framework using the PROMETHEE method to optimize client selection in Federated Learning. It enhances convergence speed and accuracy, improving FL efficiency in resource-constrained environments.