AI Paper Atlas

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

FedFed: Feature Distillation against Data Heterogeneity in Federated Learning

Zhiqin Yang, Yonggang Zhang, Yu Zheng, Xinmei Tian, Hao Peng, Tongliang Liu, Bo Han

NeurIPS 2023 · Federated learning and privacy › Personalized and heterogeneous FL · approach: Non-IID federated learning · 27 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 FedFed shared performance-sensitive features to FedAvgFedAvg · CIFAR-10 alpha=0.1, K=100top-1 accuracy: 49.72 -> 84.06Improved
add FedFed to FedNovaFedNova · CIFAR-10 alpha=0.1, K=100top-1 accuracy: 45.97 -> 86.64Improved
add FedFed to SCAFFOLDSCAFFOLD · CIFAR-10 alpha=0.05, K=10top-1 accuracy: 37.87 -> 81.04Improved
add FedFed to FedAvgFedAvg · CIFAR-100 alpha=0.1, E=1, K=10top-1 accuracy: 67.84 -> 69.64Improved
add FedFed to FedNovaFedNova · CIFAR-100 alpha=0.1, E=5, K=10top-1 accuracy: 69.78 -> 69.98No clear effect
add FedFed to FedAvgFedAvg · SVHN alpha=0.1, E=1, K=100communication rounds to target: 623 -> 763 roundsMixed
add FedFed to FedAvgFedAvg · CIFAR-10 #C=2 partitiontest accuracy: 42.54 -> 89.23Improved
use Laplacian instead of Gaussian DP noise (ablation)FedFed · CIFAR-10test accuracy: 92.34 -> 92.30 (FedAvg)No clear effect
increase DP noise level sigma_s^2 (ablation)FedFed · FMNISTtest accuracyHurt
decrease constraint rho (ablation)FedFed · CIFAR-10global model accuracyHurt
use beta-VAE instead of ResNet generator for feature distillation (ablation)FedFed · CIFAR-10convergence/accuracyImproved
share full data with DP noise instead of only performance-sensitive features (ablation)FedFed · FL system accuracytest accuracyHurt

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