AI Paper Atlas

Research map / Federated learning and privacy / Differential privacy

DP-KFC: Data-Free Preconditioning for Privacy-Preserving Deep Learning

Marc Molina Van den bosch, Riccardo Taiello, Albert Sund Aillet, Andrea Protani, Miguel Angel Gonzalez Ballester, Luigi Serio

ICML 2026 · Federated learning and privacy › Differential privacy · approach: Private training with public data · 0 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 synthetic-noise KFAC preconditioning instead of plain DP-SGDDP-SGD · MNIST CNN (epsilon=1)test accuracy (%): 91.7 -> 94.2Improved
use synthetic-noise KFAC preconditioning instead of DP-AdamDP-Adam on BERT · StackOverflow (epsilon=1)accuracy (%): 89.5 -> 91.8Improved
use synthetic-noise KFAC preconditioning instead of DP-SGDDP-SGD logistic regression · IMDB (epsilon=2.8)accuracy (%): 83.1 -> 85.8Improved
use synthetic probes instead of public data for KFAC factorsPublic DP-KFC · StackOverflow BERT (epsilon=1)accuracy (%): 96.1 -> 91.8Hurt
use synthetic pink noise instead of MNIST proxy dataPublic DP-KFC · PathMNIST<-MNIST transfer (epsilon=1)test accuracy (%): 73.4 -> 78.2Improved
use synthetic pink noise instead of MNIST proxy dataPublic DP-KFC · Fashion<-MNIST transfer (epsilon=1)test accuracy (%): 87.6 -> 87.8No clear effect
replace public proxy with pink noise in AdaDPSAdaDPS · PathMNIST<-MNIST transfer (epsilon=1)test accuracy (%): 70.5 -> 71.2No clear effect
add block-diagonal KFAC geometry preconditioning rather than diagonal AdaDPSAdaDPS · MNIST CNN (epsilon=1)test accuracy (%): 91.3 -> 94.2Improved
combine Synthetic DP-KFC with post-privatization DP-AdamBCSynthetic DP-KFC · MNIST CNN (epsilon=1)test accuracy (%): 94.2 -> 95.5Improved
use synthetic DP-KFC instead of DP-AdamBC aloneDP-AdamBC · MNIST CNN (epsilon=1)test accuracy (%): 94.0 -> 94.2Improved
train frozen-backbone CrossViT with DP-KFC rather than DP-AdamDP-Adam · CIFAR-100 CrossViTtest accuracy (%): about +1.4%Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 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