AI Paper Atlas

Research map / Federated learning and privacy / Differential privacy

Beyond Membership: Limitations of Add/Remove Adjacency in Differential Privacy

Gauri Pradhan, Joonas Jälkö, Santiago Zanella-Beguelin, Antti Honkela

ICLR 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
audit with substitute-adjacency worst-case dataset canaries instead of relying on add/remove accountingDP-SGD · worst-case crafted datasetempirical epsilonImproved
use crafted gradient canariesDP-SGD fine-tuning ViT-B-16 · CIFAR10empirical epsilonImproved
use input-space canariesDP-SGD fine-tuning ViT-B-16 · CIFAR10empirical epsilonMixed
lower the subsampling rate q (ablation)input-space canaries · CIFAR10 fine-tuningaudit efficacyHurt
use input-space canaries on a model trained from scratchDP-Adam MLP · Purchase100empirical epsilonHurt
use crafted gradient canaries on a model trained from scratchDP-Adam MLP · Purchase100empirical epsilonImproved
use gradient canariesDP-SGD Sentence-BERT linear layer · SST-2empirical epsilonImproved
add the canary to the training dataDP-SGD models · CIFAR10, Purchase100, SST-2test accuracyNo clear effect

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