Federated learning (FL) is an emerging distributed machine learning paradigm enabling collaborative model training on decentralized devices without exposing their local data. A key challenge in FL is the uneven data distribution across client devices, violating the well-known assumption of independent-and-identically-distributed (IID) training samples in conventional machine learning. Clustered federated learning (CFL) addresses this challenge by grouping clients based on the similarity of their data distributions. However, existing CFL approaches require a large number of communication rounds for stable cluster formation and rely on a predefined number of clusters, thus limiting their flexibility and adaptability. This paper proposes FedClust, a novel CFL approach leveraging correlations between local model weights and client data distributions. FedClust groups clients into clusters in a one-shot manner using strategically selected partial model weights and dynamically accommodates newcomers in real-time. Experimental results demonstrate FedClust outperforms baseline approaches in terms of accuracy and communication costs.
Federated Learning (FL) is an emerging paradigm in machine learning without exposing clients' raw data. In practical scenarios with numerous clients, encouraging fair and efficient client participation in federated learning is of utmost importance, which is also challenging given the heterogeneity in data distribution and device properties. Existing works have proposed different client-selection methods that consider fairness; however, they fail to select clients with high utilities while simultaneously achieving fair accuracy levels. In this paper, we propose a fair client-selection approach that unlocks threefold fairness in federated learning. In addition to having a fair client-selection strategy, we enforce an equitable number of rounds for client participation and ensure a fair accuracy distribution over the clients. The experimental results demonstrate that FedFair^3, in comparison to the state-of-the-art baselines, achieves 18.15% less accuracy variance on the IID data and 54.78% on the non-IID data, without decreasing the global accuracy. Furthermore, it shows 24.36% less wall-clock training time on average.