AI Paper Atlas

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

Auto-FedRL: Federated Hyperparameter Optimization for Multi-Institutional Medical Image Segmentation

Pengfei Guo, Dong Yang, Ali Hatamizadeh, An Xu, Ziyue Xu, Wenqi Li, Can Zhao, Daguang Xu, Stephanie Harmon, Evrim Turkbey, Baris Turkbey, Bradford Wood and 5 more

ECCV 2022 · Federated learning and privacy › Personalized and heterogeneous FL · approach: Federated medical and multimodal learning

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 local epochs to the search space (ablation)Auto-FedRL(CS) · CIFAR-10 (8 non-IID clients)accuracy (%): 90.02 -> 90.10Improved
add aggregation weights to the search space (ablation)Auto-FedRL(CS) · CIFAR-10 (8 non-IID clients)accuracy (%): 90.10 -> 90.52Improved
add server learning rate to the search space (ablation)Auto-FedRL(CS) · CIFAR-10 (8 non-IID clients)accuracy (%): 90.52 -> 90.85Improved
add all four hyperparameters vs learning rate alone (ablation)Auto-FedRL(MLP) · CIFAR-10 (8 non-IID clients)accuracy (%): 90.12 -> 91.27Improved
add local epochs to search with discrete search (ablation)Auto-FedRL(DS) · CIFAR-10 (8 non-IID clients)accuracy (%): 89.83 -> 89.86No clear effect
replace discrete with continuous search (ablation)Auto-FedRL · CIFAR-10 (8 non-IID clients)search memory and time: 42.8 GB -> 3.00 GB; 8.246 s -> 0.012 sImproved
replace discrete with continuous search (ablation)Auto-FedRL · CIFAR-10 (8 non-IID clients)accuracy (%): 90.70 -> 90.85Improved
use an MLP RL agent instead of direct Gaussian update (ablation)Auto-FedRL(CS) · CIFAR-10 (8 non-IID clients)accuracy (%): 90.85 -> 91.27Improved
use an MLP RL agent instead of direct Gaussian update (ablation)Auto-FedRL(CS) · COVID-19 lesion segmentationDice global test avg (%): 64.7 -> 64.0Hurt
use an MLP RL agent instead of direct Gaussian update (ablation)Auto-FedRL(CS) · Pancreas segmentationDice global test avg (%): 76.5 -> 76.6No clear effect
use Auto-FedRL(CS) instead of Auto-FedAvgFL baseline · COVID-19 lesion segmentationDice global test avg (%): 63.5 -> 64.7Improved
use Auto-FedRL(MLP) instead of best baselineFL baseline · Pancreas segmentationDice global test avg (%): 75.3 -> 76.6Improved

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