AI Paper Atlas

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

Personalized Federated Learning under Local Supervision

Qiqi Liu, Jiaqiang Li, Yuchen Liu, Yaochu Jin, Lingjuan Lyu, Xiaohu Wu, Han Yu

ICCV 2025 · Federated learning and privacy › Personalized and heterogeneous FL · approach: Personalization with global-local trade-off · 4 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
let clients use heterogeneous supervisor architectures (Transformer, CNN, LeNet-5) (ablation)FedSimSup · CIFAR10 (Dirichlet 0.1, 100 clients)accuracy: .892 -> .861Hurt
let clients use heterogeneous supervisors (ablation)FedSimSup · CIFAR100 (Dirichlet 0.1, 100 clients)accuracy: .503 -> .460Hurt
use pre-trained Transformer supervisor on a client (ablation)FedSimSup · CIFAR10 (Dirichlet 0.1, 100 clients)accuracy: .892 -> .898Improved
use CNN supervisor instead of LeNet-5 supervisor (ablation)FedSimSup-TCL · CIFAR10 (Dirichlet 0.1, 100 clients)accuracy: .851 -> .835Hurt
add similarity information for model aggregation (ablation)FedSimSup · CIFAR10 (Dirichlet 0.1)convergence speedImproved
use parallel supervisor architecture instead of serial FedRep-style architecture (ablation)FedSimSup · CIFAR100 (Dirichlet 0.1)accuracyImproved
increase C and gamma to (40, 3/7) from smaller pairs (ablation)FedSimSup · CIFAR10 and CIFAR100accuracyImproved
add FedSimSup supervisor and similarity aggregation versus FedAsFedAs · CIFAR100 (Dirichlet 0.1, 100 clients)accuracy: .467 -> .503Improved
add FedSimSup versus local training on Pathological splitLocal · CIFAR100 (Pathological, 100 clients, 20 shards)accuracy: .275 -> .464Improved
apply FedSimSup under Pathological split with 5 shardspFedFda · CIFAR100 (Pathological, 100 clients, 5 shards)accuracy: .642 -> .637Hurt

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