Temporal membership inference attacks on continually updated models
TMIA black-box attack infers whether and when samples were used in training for language and image models
Stories per day · 30 days
1 RSS
Relateddifferential-privacy2model-privacy2privacy2bpe1continual-learning1diffusion-models1dnn1dp-sgd1gaussian1language-models1mean-estimation1medical-records1model-auditing1olmo-21on-policy1qnn1quantization1reconstruction-loss1
17 stories, newest first · one card per story
Stories tagged #membership-inference
OLMo 2 membership inference study finds limited success
Controlled OLMo 2 1B training with matched document pairs shows reference-ratio AUC of 0.767 but detects only 3.8% members at 1% FPR.
Paper calibrates diffusion membership inference across two time axes
Reference-Model Calibration uses an earlier checkpoint; Temporal Self-Calibration uses one model's neighboring diffusion timesteps.
BPE tokenizers can leak whether a dataset was in training
RTAM-Fusion uses rare-token mass and fewer shadow tokenizers to beat Vocabulary Overlap on a membership benchmark.
4-8 bit quantized DNNs stay exposed to membership inference
Eight attacks across six quantization methods find privacy gains mainly at 2 bits, with large utility loss.
Paper restates robust membership inference for mean estimation
The attacks remain effective when released statistics are highly noisy, for Gaussians and binary products.
OP-Print targets membership inference in on-policy self-distillation
Gray-box and black-box signals use sparse correction imprints, while the authors say baseline attacks stay near chance.
Study compares privacy and capability of synthetic medical-record training
Direct finetuning leaked most; DP-SGD cut leakage but hurt capability; privacy-aware rewriting balanced both better.
Efficient Provably Private Classification with a Tabular Foundation Model
PrivTab is a differentially private tabular foundation model that classifies sensitive data in one forward pass.
Closed-Form Noise Calibration Against Membership Inference for Random-Allocation DP-SGD
A closed-form noise formula bounds membership-inference accuracy for random-allocation DP-SGD and needs less noise.
TranScope: What the Software Hides About LLM Training Data, the Hardware Reveals at Scale, and Accelerators Magnify
TranScope detects LLM training membership from TLB and accelerator timing without a surrogate model.
Near-Duplicate Families Break Exact-Record Membership Inference
Near-duplicates make LiRA membership inference mark 99.7% of exact non-members as members on CC-News.
Leaky Students: Membership Inference against On-Policy Distillation
Researchers show on-policy distillation students leak teacher-record membership, with Leaky reaching mean AUROC 0.875.
Don't Send What You Don't Need: Question-Guided Token Pruning as a Privacy Defense for Vision-Language Models
QPriv-VL prunes privacy-sensitive visual tokens in federated/split VQA, cutting membership-inference success on VQA-RAD from 0.99 to 0.76-0.79 using ~40% of tokens.
Predicting Privacy Leakage from Weight Spectral Density
Study shows WeightWatcher spectral metrics like stable rank correlate with membership inference vulnerability, enabling cheaper ML privacy auditing.
Subgroup Membership Inference Audits of Differentially Private Synthetic Text
Audits of 32 differentially private synthetic-text releases show subgroup membership leakage is concentrated in few records and systematically underestimated by average-case attacks.
Client-Side Probing of Deleted Ridge Statistics in Federated Unlearning
Researchers show malicious federated learning clients can probe broadcast classifiers to recover deleted samples, exposing exact label leakage on MNIST and CIFAR-10.