Federated Learning (FL) is an emerging machine learning paradigm that allows multiple end-users to collaboratively train models without sharing their private data. Existing FL frameworks are undergoing a significant change fueled by Foundation Models (FM), which presents a unique opportunity to unlock new possibilities, challenges, and applications in AI research. FL can be a beneficial tool to address the shortage of high-quality legalized data required by FM training. Meanwhile, FM can empower existing FL systems to alleviate performance degradation problems and lead to a better balance between generalization and personalization, diversity and fidelity. A robust, trustworthy FL platform can be established by examining the interplay between FL and FM, allowing them to benefit each other mutually. Also, it is necessary to overcome potential challenges incurred by original heterogeneity issues, high communication and computation costs, and privacy and security issues in FL.

This special session on trustworthy FL aims to explore recent advances in the intersection of the Foundation Model (FM) and Federated Learning (FL), inspiring future research that can enhance both fields and propel the development of trustworthy AI systems. We invite papers to present novel ideas, showcase potential applications, and discuss promising directions in this research field. We welcome contributions on all aspects of trustworthy FL, with a special focus on its intersection with foundation models. Topics include but are not limited to:

Algorithmic Advances, Novel Issues, and Open Challenges in FL
Theoretical Analysis for Trustworthy FL
Advancing Trustworthy FL for FM
Robustness and Reliability for Trustworthy FL
Improving FM with Decentralized Data
Heterogeneity in FL and FM
Fairness of FL and FM
Interpretability and Explainability of FL and FM
Personalized FL based on FM
FL Trust Policy and Strategies
Performance Evaluation Methods, Metrics of FL Systems
Tools and Resources (e.g., Benchmark Datasets, Software Libraries, ...)
Security and Privacy of FL for FM (e.g., Differential Privacy, Adversarial Attacks, Poisoning Attacks, Inference Attacks, Data Anonymization, Model Distillation, Secure Multi-Party Computation, etc.)
Applications of FL (e.g., Healthcare, Edge Devices, Advertising, Social Networks, Blockchain, Web Search, etc.)


Submission Instructions:
Information on paper submission can be found here: https://2024.ieeewcci.org/submission

All accepted papers will be included in the WCCI-2024 proceedings, published on the IEEE Xplore Digital Library.


Organizers:
Guodong Long (University of Technology Sydney, Australia)
Zenglin Xu (Harbin Institute of Technology, China)
Jing Jiang (University of Technology Sydney, Australia)
Yue Tan (University of Technology Sydney, Australia)
Han Yu (Nanyang Technological University, Singapore)
Irwin King (The Chinese University of Hong Kong, Hong Kong)