AI Paper Atlas

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

Federated Domain Generalization with Domain-specific Soft Prompts Generation

Jianhan Wu, Xiaoyang Qu, Zhangcheng Huang, Jianzong Wang

ICCV 2025 · Federated learning and privacy › Personalized and heterogeneous FL · approach: Federated prompt and LoRA tuning · 2 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
add the second-step generative prompt model (GAN) to DSPs (ablation)FedDSPG · Office-Homeaccuracy: 82.5 -> 85.3Improved
use domain-specific soft prompts instead of conditional soft prompts (ablation)FedDSPG · Office-Homeaccuracy: 84.6 -> 85.3Improved
use learnable soft prompts instead of hand-crafted prompts (ablation)FedDSPG · Office-Homeaccuracy: 77.7 -> 85.3Improved
add domain-specific prompts without generative model (ablation)ADAPT-like (Ours-WGM) · Office-Homeaccuracy: 82.9 -> 82.5No clear effect
add momentum-based aggregation (ablation)FedDSPG · Office-Homeaccuracy: 84.5 -> 85.3Improved
set momentum coefficient alpha to 0.2 instead of 1.0 (ablation)FedDSPG · Office-Homeaccuracy: 83.39 -> 85.30Improved
communicate every 2 epochs instead of 1 (ablation)FedDSPG · Office-Homeaccuracy: 85.3 -> 84.6Hurt
communicate every 0.5 epochs instead of 1 (ablation)FedDSPG · Office-Homeaccuracy: 85.3 -> 85.3No clear effect
use FedDSPG instead of FedTPGFedTPG · DomainNetaccuracy: 72.7 -> 75.6Improved
use FedDSPG instead of FedTPGFedTPG · DomainNet to Office-Home transferaccuracy: 75.6 -> 79.8Improved
use FedDSPG instead of FedTPGFedTPG · DomainNet to PACS transferaccuracy: 90.7 -> 93.0Improved
use FedDSPG instead of ADAPTADAPT · DomainNet (Clp target)accuracy: 77.5 -> 74.7Hurt

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