AI Paper Atlas

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

FedAU2: Attribute Unlearning for User-Level Federated Recommender Systems with Adaptive and Robust Adversarial Training

Yuyuan Li, Junjie Fang, Fengyuan Yu, Xichun Sheng, Tianyu Du, Xuyang Teng, Shaowei Jiang, Linbo Jiang, Jianan Lin, Chaochao Chen

AAAI 2026 · Federated learning and privacy › Privacy attacks and secure learning · approach: Defending federated learning against poisoning · 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 FedAU2 attribute unlearning instead of no unlearningFedNCF and FedVAE · ML-1M, ML-100K, LastFMBAcc: average BAcc reduction 26.42% vs 11.5% for APMImproved
use FedAU2 on a personalized modelFedRAP · three datasetsBAcc: 14.09% vs 9.24% for APMImproved
FedAU2 versus APM on recommendation qualityFedRecs · three datasetsNDCG@10: -4.51% (FedAU2) vs -20.05% (APM)Improved
add SUT instead of Global adversarial training (ablation)FedAU2 · ML-1M (gender)NDCG@10Improved
use SUT instead of Pretrain for adversarial model (ablation)FedAU2 · ML-1M (gender)BAccImproved
use DSVAE instead of original MLP adversary (ablation)FedVAE · ML-1M (gender), gradient attackattack accuracy: 90% -> 58%Improved
use plain VAE instead of original MLP adversary (ablation)FedVAE · ML-1M (gender), gradient attackattack accuracy: 90% -> 89%No clear effect
increase stochasticity coefficient lambda (ablation)DSVAE · ML-1M (gender), FedVAEBAcc / NDCG@10Mixed
FedAU2 on FedRAP recommendationFedRAP · three datasetsHR@10/NDCG@10: ML-1M gender NDCG@10 0.2381 -> 0.2437No 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