Adversarial perturbations reliably fabricate membership signals in vision-model MIAs, separated by a gradient-norm collapse trajectory that enables robust detection and inference.
arXiv preprint arXiv:2111.08440 , year=
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
citation-role summary
method 1
citation-polarity summary
roles
method 1polarities
use method 1representative citing papers
Entity-level membership inference determines whether information about a target real-world entity was used in LLM training, using only black-box generated text and achieving AUC up to 0.97 on person entities.
Finetuning open LMs on ChatGPT outputs creates models that mimic style and fool human raters but fail to close the performance gap to proprietary systems on tasks not well-represented in the imitation data.
Advanced generalization techniques such as augmentation and early stopping can reduce membership inference attack success rates by up to 100 times, confirmed across more than 1,000 models.
citing papers explorer
-
Have You Ever Seen Them? Entity-level Membership Inference through Interrogating Large Language Models
Entity-level membership inference determines whether information about a target real-world entity was used in LLM training, using only black-box generated text and achieving AUC up to 0.97 on person entities.
-
Generalization and Membership Inference Attack a Practical Perspective
Advanced generalization techniques such as augmentation and early stopping can reduce membership inference attack success rates by up to 100 times, confirmed across more than 1,000 models.