AI Paper Atlas

Research map / Federated learning and privacy / Differential privacy

(Amplified) Banded Matrix Factorization: A unified approach to private training

Christopher A. Choquette-Choo, Arun Ganesh, Ryan McKenna, Hugh Brendan McMahan, J Keith Rush, Abhradeep Guha Thakurta, Zheng Xu

NeurIPS 2023 · Federated learning and privacy › Differential privacy · approach: Differentially private federated learning · 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
use banded matrix (b=342) instead of DP-SGD (1 band) (ablation)matrix mechanism · prefix sums, n=2052RMSE under b>=342-min-sep: 9.63 -> 1.05Improved
use banded matrix (b=342) instead of Optimal TreeAggmatrix mechanism · prefix sums, n=2052RMSE under b>=342-min-sep: 1.53 -> 1.05Improved
reduce bands from 342 to 128 (ablation)BandMF · prefix sums, n=2052RMSE under b>=342-min-sep: 1.05 -> 1.27Hurt
add equal column norm constraint (ablation)MF (b=342) · prefix sums, n=2052RMSE under b>=342-min-sep: 1.08 -> 1.05Improved
add equal column norm constraint (ablation)MF (b=342) · prefix sums, n=2052RMSE under (6,342)-participation: 1.04 -> 1.05Hurt
switch from (k,b)-participation to b-min-sep-participationMF without column normalization · prefix sumsRMSE: 1.00 -> 1.05Mixed
BandMF with amplification vs prior MFBandMF · CIFAR-10, eps in [2,5]accuracy: ~1 percentage pointImproved
BandMF with amplification vs DP-SGDBandMF · CIFAR-10, eps below 2accuracy: 3 percentage pointsImproved
BandMF vs DP-SGD as eps becomes very smallBandMF · CIFAR-10accuracyNo clear effect
combine amplification with MFBandMF · StackOverflow NWP, eps in [1,8]accuracyImproved
use BandMF instead of Multi-epoch MF at eps=16BandMF · StackOverflow NWPaccuracyNo clear effect
use BandMF instead of Online TreeAggBandMF · production Spanish NWP FLprivacy guarantee: 0.52 -> 0.24Improved

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