AI Paper Atlas

Research map / Federated learning and privacy / Differential privacy

On Privacy and Personalization in Cross-Silo Federated Learning

Ken Liu, Shengyuan Hu, Steven Wu, Virginia Smith

NeurIPS 2022 · Federated learning and privacy › Differential privacy · approach: Differentially private federated learning · 20 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
Use MR-MTL instead of the benchmark methodsPrivate cross-silo FL · Vehicle, School, GLEAM, and CIFAR-10utilityImproved
Use MR-MTL instead of local training or FedAvgPrivate cross-silo FL · Vehicle, School, and GLEAMutilityImproved
Increase MR-MTL's regularization strength λ under stronger privacyMR-MTL · Mean estimationgeneralization errorMixed
Warm-start MR-MTL with private IFCA clustering, then train within formed clustersMR-MTL · Rotated & Masked MNISTtest accuracyImproved
Increase silo-specific heterogeneity after warm-starting MR-MTL with IFCAMR-MTL with IFCA warm-start · Rotated & Masked MNISTtest accuracyMixed
Use local finetuning after shared federated training rather than training locally throughoutLocal finetuning · Private cross-silo FLutilityNo clear effect
Run IFCA cluster selection at every roundIFCA · Private cross-silo FLutilityHurt
Account for Ditto's extra local training iterations under a fixed privacy budgetDitto · Private cross-silo FLutilityHurt
Privately tune MR-MTL's λ using a moderate number of trialsMR-MTL · Mean estimationutilityHurt

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