AI Paper Atlas

Research map / Federated learning and privacy / Personalized and heterogeneous FL

Prototype Normalization: Optimizing Prototype Separation for Heterogeneous Federated Learning

Daeyoung Choi, Jihwan Shin, Gyuejeong Lee

ECCV 2026 · Federated learning and privacy › Personalized and heterogeneous FL · approach: Non-IID federated learning · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (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 changedWhereMeasuredEffect
replace FedProto weighted averaging with ProtoNorm hyperspherical alignment plus upscalingFedProto · CIFAR-100 practical alpha=0.1test accuracy (%): 41.31 -> 47.94Improved
ProtoNorm vs FedTGPprototype-based FL · Tiny ImageNet pathologicaltest accuracy (%): 31.85 -> 40.24Improved
ProtoNorm vs FedKD on Tiny ImageNet practicalheterogeneous FL · Tiny ImageNet practical alpha=0.1 (large-scale models)test accuracy (%): 28.87 -> 29.97Improved
ProtoNorm vs LG-FedAvg/FedKD on CIFAR-10 practicalheterogeneous FL · CIFAR-10 practical alpha=0.1test accuracy (%): 88.10 -> 88.95Improved
ProtoNorm vs FedTGP at 100 clients with 20% samplingprototype-based FL · CIFAR-100 (100 clients, 20% sampling)test accuracy (%): 39.20 -> 41.30Improved
ProtoNorm vs FedTGP across Dirichlet alpha 0.01prototype-based FL · CIFAR-100 (Dirichlet alpha=0.01)test accuracy (%): 72.67 -> 72.95Improved
ProtoNorm vs FedTGP on ImageNet-1K 32x32prototype-based FL · ImageNet-1K 32x32 (10 clients)test accuracy (%): 13.62 -> 21.52Improved
ProtoNorm vs FedUV in homogeneous settinghyperspherical FL · CIFAR-100 (homogeneous ResNet-18)test accuracy (%): 48.33 -> 52.67Improved
add prototype alignment and upscaling together (ablation)FedProto · Flowers-102 (ablation)test accuracy (%): 41.22 -> 57.52Improved
add prototype alignment alone (unit-norm global prototypes) (ablation)FedProto · Flowers-102 (ablation)test accuracy (%): 41.22 -> 37.32Hurt
add prototype alignment alone (ablation)FedProto · Tiny ImageNet (ablation)test accuracy (%): 17.88 -> 18.26No clear effect
add prototype upscaling alone (ablation)FedProto · CIFAR-100 (ablation)test accuracy (%): 41.31 -> 40.26No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy