Search papers, labs, and topics across Lattice.
The paper introduces Cluster-Guided FedRec (CGFedRec), a federated recommendation framework that uses item clusters as structural constraints to guide local item representation learning. CGFedRec addresses the communication bottleneck in federated recommendation by transmitting compact cluster labels instead of full item embeddings, thereby reducing communication costs. Experiments on multiple datasets demonstrate that CGFedRec achieves superior recommendation accuracy and communication efficiency compared to methods relying on synchronizing high-dimensional item representations.
Ditch the heavy embedding sync: this federated recommendation method slashes communication costs by transmitting lightweight cluster labels instead of full item embeddings, without sacrificing accuracy.
Federated recommendation facilitates collaborative model training across distributed clients while keeping sensitive user interaction data local. Conventional approaches typically rely on synchronizing high-dimensional item representations between the server and clients. This paradigm implicitly assumes that precise geometric alignment of embedding coordinates is necessary for collaboration across clients. We posit that establishing relative semantic relationships among items is more effective than enforcing shared representations. Specifically, global semantic relations serve as structural constraints for items. Within these constraints, the framework allows item representations to vary locally on each client, which flexibility enables the model to capture fine-grained user personalization while maintaining global consistency. To this end, we propose Cluster-Guided FedRec framework (CGFedRec), a framework that transforms uploaded embeddings into compact cluster labels. In this framework, the server functions as a global structure discoverer to learn item clusters and distributes only the resulting labels. This mechanism explicitly cuts off the downstream transmission of item embeddings, relieving clients from maintaining global shared item embeddings. Consequently, CGFedRec achieves the effective injection of global collaborative signals into local item representations without transmitting full embeddings. Extensive experiments demonstrate that our approach significantly improves communication efficiency while maintaining superior recommendation accuracy across multiple datasets.
