AI Paper Atlas

Research map / Federated learning and privacy / Privacy attacks and secure learning

Sibai: A Few-Shot Meta-Classifier for Poisoning Detection in Federated Learning

Melanie Götz, Torsten Krauß, Alexandra Dmitrienko

ICCV 2025 · Federated learning and privacy › Privacy attacks and secure learning · approach: Defending federated learning against poisoning · 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
Sibai instead of MESASFL poisoning detection · Scenario 4 (0.9 1-class non-i.i.d.), random label flipF1: 0.0000 -> 0.8235Improved
Sibai instead of MESASFL poisoning detection · Scenario 1 (i.i.d. low PDR), blend backdoorF1: 0.1818 -> 1.0000Improved
Sibai instead of MESASFL poisoning detection · Scenario 6 (inter-client non-i.i.d.), semantic backdoorF1: 0.3157 -> 0.9473Improved
Sibai instead of HDBSCANFL poisoning detection · Scenario 11 (adaptive), pixel backdoorF1: 1.0000 -> 0.9411Hurt
Sibai instead of MESASFL poisoning detection · Scenario 11 (adaptive), pixel backdoorF1: 0.6666 -> 0.9411Improved
lower poison data rate (high to low PDR) vs SibaiSibai · i.i.d. CIFAR-10, semantic backdoorF1: 1.0000 -> 0.8750Hurt
increase 1-class non-i.i.d. degree q from 0.6 to 0.9Sibai · random label flip, high PDRF1: 1.0000 -> 0.8235Hurt
add cosine and Euclidean distance features to PCA input (ablation)Sibai · detection accuracydetection accuracy: up to 10%Improved
attacker adapts to cosine distance (alpha 0.3)Sibai · i.i.d. scenarios 8 and 9F1: F1 1.0000 for all attacksNo clear effect
attacker adapts to Euclidean distance (alpha 0.3), low PDRSibai · scenario 11, pixel and blendF1: F1 1.0000 -> 0.9411Mixed
use Sibai instead of FLTrustFL defense · Scenario 4 pixel backdoorbackdoor accuracy (BA, lower better): BA 53.1 -> 0.0Improved
use Sibai instead of KrumFL defense · Scenario 1 random label flipMA diff: MA diff -35.0 -> -0.8Improved

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