Representation-Aligned Multi-Scale Personalization for Federated Learning

2026-04-13

AI Summary

This work addresses the challenge in federated learning of simultaneously accommodating client resource heterogeneity and model personalization, which is often hindered by conventional approaches relying on a fixed global backbone network that limits architectural diversity and representational adaptability. To overcome this limitation, the authors propose FRAMP, a framework that abandons the static global model in favor of dynamically generating compact, client-specific subnetworks from lightweight client descriptors. FRAMP integrates a cross-client representation alignment mechanism to preserve global semantic consistency while adapting to local data distributions and computational budgets. By unifying resource-adaptive neural architecture search with federated optimization, the method enables fine-grained, multi-scale personalized modeling. Empirical evaluations demonstrate that FRAMP significantly improves generalization and adaptability across both vision and graph benchmarks, effectively supporting efficient federated learning in highly heterogeneous environments.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Search and Optimization: Learning to Search
• Multiagent Systems: Adversarial Agents

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

In federated learning (FL), accommodating clients with diverse resource constraints remains a significant challenge. A widely adopted approach is to use a shared full-size model, from which each client extracts a submodel aligned with its computational budget. However, regardless of the specific scoring strategy, these methods rely on the same global backbone, limiting both structural diversity and representational adaptation across clients. This paper presents FRAMP, a unified framework for personalized and resource-adaptive federated learning. Instead of relying on a fixed global model, FRAMP generates client-specific models from compact client descriptors, enabling fine-grained adaptation to both data characteristics and computational budgets. Each client trains a tailored lightweight submodel and aligns its learned representation with others to maintain global semantic consistency. Extensive experiments on vision and graph benchmarks demonstrate that FRAMP enhances generalization and adaptivity across a wide range of client settings.

Problem

• federated learning
• resource constraints
• personalization
• model diversity
• representation alignment

Innovation

• federated learning
• personalization
• resource-adaptive
• representation alignment
• multi-scale