AI Paper Atlas

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

A Framework for Double-Blind Federated Adaptation of Foundation Models

Nurbek Tastan, Karthik Nandakumar

ICCV 2025 · Federated learning and privacy › Personalized and heterogeneous FL · approach: Federated learning with pretrained models · 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
replace linear probing with BlindFed (parallel adapter)ViT-Base · SVHN (centralized, pooled)balanced accuracy: 0.5938 -> 0.9232Improved
replace linear probing with BlindFedViT-Base · CIFAR-100 (federated K=5, Dirichlet 0.01)balanced accuracy: 0.5317 -> 0.6620Improved
replace linear probing with BlindFedViT-Base · CIFAR-10 (federated K=50, Dirichlet 1.0)balanced accuracy: 0.9007 -> 0.9287Improved
add stochastic block sampling (SBS) to BlindFedBlindFed · CIFAR-10 (centralized, pooled)balanced accuracy: 0.9428 -> 0.9443No clear effect
add SBS to BlindFedBlindFed · CIFAR-100 (centralized, pooled)balanced accuracy: 0.7930 -> 0.7869No clear effect
add SBS to BlindFedBlindFed · CIFAR-10 (federated K=50, Dirichlet 1.0)balanced accuracy: 0.9287 -> 0.9388Improved
add SBS to BlindFedBlindFed · Fed-ISIC2019 (federated K=5, aux Fed-ISIC2019 center 0, InD)balanced accuracy: 0.6679 -> 0.6831Improved
add SBS to BlindFedBlindFed · CIFAR-10 (federated K=5, Dirichlet 0.01)balanced accuracy: 0.8540 -> 0.8489Hurt
use out-of-distribution Tiny-ImageNet instead of in-distribution Fed-ISIC2019 as auxiliary datasetBlindFed · Fed-ISIC2019 (federated K=5)balanced accuracy: 0.6679 -> 0.6481Hurt
compare BlindFed to LoRALoRA · Fed-ISIC2019 (federated K=5, aux InD)balanced accuracy: 0.6844 -> 0.6831Mixed
raise softmax Taylor degree from 2 to 6 (ablation)FHE-friendly ViT · Fed-ISIC2019 softmax approximationone-epoch accuracy (%): 75.17 -> 84.36Improved
raise softmax Taylor degree from 6 to 16 (ablation)FHE-friendly ViT · Fed-ISIC2019 softmax approximationone-epoch accuracy (%): 84.36 -> 85.08No clear effect

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