FedDAG: Clustered Federated Learning via Global Data and Gradient Integration for Heterogeneous Environments
#Federated Learning #Clustering #Data Heterogeneity #Gradient Similarity #Dual Encoder #Machine Learning #Distributed Computing
📌 Key Takeaways
- FedDAG introduces a weighted, class‑wise similarity metric that jointly considers data and gradient information, offering a more holistic measure of client similarity. The framework uses a dual‑encoder architecture: a primary encoder trained on a cluster’s own clients, and a secondary encoder refined using gradients from complementary clusters, enabling cross‑cluster feature transfer while preserving specialty. Unlike prior clustered federated learning methods that limit knowledge sharing to within a single cluster, FedDAG facilitates cross‑cluster collaboration, mitigating the divide between diverse client populations. Experimental results across diverse benchmark datasets and data‑heterogeneity regimes demonstrate that FedDAG consistently outperforms existing clustered federated learning baselines in terms of predictive accuracy.
📖 Full Retelling
A new federated learning framework called FedDAG – short for Clustered Federated Learning via Global Data and Gradient Integration – has been presented by Anik Pramanik, Murat Kantarcioglu, Vincent Oria, and Shantanu Sharma. The paper, submitted on 26 February 2026 and available on arXiv (2602.23504), tackles the problem that conventional federated learning performs poorly when client data are heterogeneous. It introduces a dual‑encoder clustered approach that integrates both data and gradient similarity into a weighted, class‑wise metric, and allows cross‑cluster feature transfer while preserving cluster‑specific specialization. In extensive experiments on a range of benchmark datasets and heterogeneity settings, FedDAG consistently outperforms state‑of‑the‑art clustered federated learning baselines in accuracy.
🏷️ Themes
Federated Learning, Client Data Heterogeneity, Clustered Federated Learning, Cross‑cluster Knowledge Transfer, Gradient‑Based Similarity, Dual‑Encoder Architecture
Entity Intersection Graph
No entity connections available yet for this article.
Original Source
--> Computer Science > Machine Learning arXiv:2602.23504 [Submitted on 26 Feb 2026] Title: FedDAG: Clustered Federated Learning via Global Data and Gradient Integration for Heterogeneous Environments Authors: Anik Pramanik , Murat Kantarcioglu , Vincent Oria , Shantanu Sharma View a PDF of the paper titled FedDAG: Clustered Federated Learning via Global Data and Gradient Integration for Heterogeneous Environments, by Anik Pramanik and 3 other authors View PDF HTML Abstract: Federated Learning enables a group of clients to collaboratively train a model without sharing individual data, but its performance drops when client data are heterogeneous. Clustered FL tackles this by grouping similar clients. However, existing clustered FL approaches rely solely on either data similarity or gradient similarity; however, this results in an incomplete assessment of client similarities. Prior clustered FL approaches also restrict knowledge and representation sharing to clients within the same cluster. This prevents cluster models from benefiting from the diverse client population across clusters. To address these limitations, FedDAG introduces a clustered FL framework, FedDAG, that employs a weighted, class-wise similarity metric that integrates both data and gradient information, providing a more holistic measure of similarity during clustering. In addition, FedDAG adopts a dual-encoder architecture for cluster models, comprising a primary encoder trained on its own clients' data and a secondary encoder refined using gradients from complementary clusters. This enables cross-cluster feature transfer while preserving cluster-specific specialization. Experiments on diverse benchmarks and data heterogeneity settings show that FedDAG consistently outperforms state-of-the-art clustered FL baselines in accuracy. Comments: This paper has been accepted in ICLR 2026 Subjects: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC) C...
Read full article at source