AI Paper Atlas

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

Feature-Aware One-Shot Federated Learning via Hierarchical Token Sequences

Shudong Liu, Hanwen Zhang, Xiuling Wang, Yuesheng Zhu, Guibo Luo

AAAI 2026 · Federated learning and privacy › Personalized and heterogeneous FL · approach: Federated learning with pretrained models · 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
add hierarchical scale encoding (HSE) (ablation)FALCON · Average over all settingsaccuracy (%): 82.32 -> 85.92Improved
add hierarchical scale encoding (HSE) (ablation)FALCON · Pneumonia (mean over alpha)accuracy (%): 82.12 -> 86.85Improved
add hierarchical scale encoding (HSE) (ablation)FALCON · PACSaccuracy (%): 96.16 -> 96.42No clear effect
add knowledge distillation (ablation)FALCON · Average over all settingsaccuracy (%): 82.82 -> 85.92Improved
add knowledge distillation (ablation)FALCON · OfficeHomeaccuracy (%): 86.87 -> 89.68Improved
use M-AR generator instead of vanilla AR (ablation)FALCON · Pneumonia (mean over alpha)accuracy (%): 72.76 -> 86.85Improved
use M-AR generator instead of non-autoregressive generator (ablation)FALCON · PMRAM (mean over alpha)accuracy (%): 78.67 -> 84.70Improved
use M-AR generator instead of AR for synthetic token fidelity (ablation)FALCON · TuberculosisMMD: 0.003868 -> 0.003227Improved
use FALCON vs best OSFL baseline on TuberculosisFedCGS · Tuberculosisaccuracy (%): 53.50 -> 87.20Improved
use FALCON vs FedCGS on Tiny-ImageNet alpha=0.1FedCGS · Tiny-ImageNet (alpha=0.1)accuracy (%): 81.53 -> 79.93Hurt
upload synthetic token sequences rather than real ones for privacyFALCON · OfficeHome reconstruction attackPSNR: 27.18 -> 13.30Improved

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