Research map / Federated learning and privacy / Personalized and heterogeneous FL
Exploiting Shared Representations for Personalized Federated Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| FedRep with many local head updates (10 local epochs) vs. more local updates | FedRep · CIFAR10 (100 clients, 2 classes/client) | average local test accuracy | Helped |
| Running more local updates in standard FedAvg | FedAvg · CIFAR10 (100 clients, 2 classes/client) | average local test accuracy | No clear effect |
| Fine-tuning the head of global FedAvg model instead of full personalization | FedAvg · CIFAR100 (100 clients, 20 classes/client) | average local test accuracy: 79.34 vs 79.15 for FedRep | Helped |
| Learning a shared low-dimensional representation with personalized heads (FedRep) vs. FedPer-style simultaneous updates | FedRep · FEMNIST (150 clients, 3 classes/client) | average local test accuracy: 78.56 vs 76.91 for FedPer | Helped |
| Replacing FedRep head minimization with 1 or 10 gradient steps (GD-GD, 10GD-GD) in linear setting (ablation) | FedRep linear representation learning · Synthetic linear regression (d=10, k=2, m=5, r=0.1) | principal angle distance convergence | Hurt |
| Fine-tuning head only on new client with FedRep representation vs Local Only | FedRep representation · FEMNIST-digits new client after training on FEMNIST-letters | local test accuracy | Helped |
| Local-only training (no federation) on CIFAR10 with 2 classes per client | Local Only · CIFAR10 (100 clients, 2 classes/client) | average local test accuracy: 89.79 vs 87.70 for FedRep | Hurt |
Closest papers by meaning
- Personalized Federated Learning via Feature Distribution Adaptation (NeurIPS 2024 · similarity 0.90)
- PeFLL: Personalized Federated Learning by Learning to Learn (ICLR 2024 · similarity 0.89)
- Tackling Feature-Classifier Mismatch in Federated Learning via Prompt-Driven Feature Transformation (NeurIPS 2025 · similarity 0.89)
- Fine-Tuning Personalization in Federated Learning to Mitigate Adversarial Clients (NeurIPS 2024 · similarity 0.88)
- Federated Model Heterogeneous Matryoshka Representation Learning (NeurIPS 2024 · similarity 0.88)
- pFedES: Generalized Proxy Feature Extractor Sharing for Model Heterogeneous Personalized Federated Learning (AAAI 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.