List your company
ResearchFormerly WizWand

Membership Inference Attack on Clinicalnotes

0.4253TPR@FPR=0.01HT-MIA
-0.0045320.1070590.218650.330241Jan 27, 2026
Updated 8mo ago

Evaluation results

MethodLinks
HT-MIATarget Model=LLaMA-3.2-1B, Fine-tuned=true2026.010.4253
RatioTarget Model=LLaMA-3.2-1B, Fine-tuned=true2026.010.3883
HT-MIATarget Model=Qwen-3-0.6B, Fine-tuned=true2026.010.3839
LossTarget Model=LLaMA-3.2-1B, Fine-tuned=true2026.010.3148
RatioTarget Model=Qwen-3-0.6B, Fine-tuned=true2026.010.2842
ZlibTarget Model=LLaMA-3.2-1B, Fine-tuned=true2026.010.2663
LossTarget Model=Qwen-3-0.6B, Fine-tuned=true2026.010.264
ZlibTarget Model=Qwen-3-0.6B, Fine-tuned=true2026.010.2407
PACTarget Model=Qwen-3-0.6B, Fine-tuned=true2026.010.2251
Min-K++Target Model=LLaMA-3.2-1B, Fine-tuned=true2026.010.2168
PACTarget Model=LLaMA-3.2-1B, Fine-tuned=true2026.010.2143
LowercaseTarget Model=LLaMA-3.2-1B, Fine-tuned=true2026.010.199
LowercaseTarget Model=Qwen-3-0.6B, Fine-tuned=true2026.010.1891
SPV-MIATarget Model=Qwen-3-0.6B, Fine-tuned=true2026.010.133
Min-K++Target Model=Qwen-3-0.6B, Fine-tuned=true2026.010.0536
HT-MIATarget Model=GPT-2, Fine-tuned=true2026.010.0524
Min-K++Target Model=GPT-2, Fine-tuned=true2026.010.0279
LossTarget Model=GPT-2, Fine-tuned=true2026.010.0223
RatioTarget Model=GPT-2, Fine-tuned=true2026.010.0201
ZlibTarget Model=GPT-2, Fine-tuned=true2026.010.0174
LowercaseTarget Model=GPT-2, Fine-tuned=true2026.010.017
SPV-MIATarget Model=GPT-2, Fine-tuned=true2026.010.015
PACTarget Model=GPT-2, Fine-tuned=true2026.010.0135
SPV-MIATarget Model=LLaMA-3.2-1B, Fine-tuned=true2026.010.012