AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Membership Inference Attack Against Large Language Model-Based Recommendation Systems: A New Distillation-Based Paradigm

Cuihong Li, Xiaowen Huang, Chuanhuan Yin, Jitao Sang

AAAI 2026 · Large language models › Language, safety and interpretability · approach: LLM pruning and scaling data · 1 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 distilled reference model with fused features instead of shadow modelMIA on LLM-based recommenders (T5) · MovieLensAUC %: 67.74 -> 90.14Improved
use distilled reference model with fused features instead of shadow modelMIA on LLM-based recommenders (T5) · Deliciousaccuracy %: 51.35 -> 66.36Improved
replace logistic regression attack model with BERT (ablation)MIA · Last.FM and MovieLens (small data)accuracy %: 65.09 -> 59.09 (Last.FM); 85.00 -> 70.00 (MovieLens)Hurt
replace logistic regression with BERT on large data (ablation)MIA · Deliciousaccuracy %: 65.27 -> 66.71, training time 16.17s -> 691.36sNo clear effect
use fused feature instead of individual features (ablation)MIA · MovieLensAUC %: 84.36 (loss only) -> 90.84Improved
use fused feature instead of loss-only on Delicious (ablation)MIA · Deliciousf1 %: 73.12 (loss only) -> 72.28Hurt
keep loss in fused feature (ablation)MIA · MovieLensAUC %: 75.42 -> 90.84Improved
keep vector feature in fused feature (ablation)MIA · DeliciousAUC %: 68.29 -> 67.55No clear effect
weight features in fusion (ablation)MIA · MovieLensaccuracy %: 80.63 -> 86.00Improved
compress MLP-lifted features back to 512D (ablation)MIA · MovieLensaccuracy %: 83.75 -> 76.25Hurt
use student with architecture matching the teacher (ablation)MIA · MovieLens (T5 teacher)AUC %: T5 student 90.42 vs GPT-2 student 63.38Improved
vary distillation weight alpha (ablation)MIA · four datasetsattack performanceMixed

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