AI Paper Atlas

Research map / Adversarial robustness and security / Attacks on text and graphs

Are You Stealing My Model? Sample Correlation for Fingerprinting Deep Neural Networks

Jiyang Guan, Jian Liang, Ran He

NeurIPS 2022 · Adversarial robustness and security › Attacks on text and graphs · approach: Membership inference and inversion · 10 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
SAC-m vs adversarial-example fingerprint CAE against adversarial extractionSAC · CIFAR10 Extract-AdvAUC: 0.52 -> 0.92Improved
SAC-w vs EWE watermark under Finetune-ASAC · Tiny-ImageNetAUC: 0.48 -> 1.00Improved
use CutMix samples instead of normal samples (ablation)SAC-m · CIFAR10 Extract-AdvAUC: 0.78 -> 0.92Improved
use CutMix samples instead of source-misclassified samples (ablation)SAC-m · CIFAR10 Extract-LAUC: 0.88 -> 1.00Improved
cosine similarity instead of Gaussian RBF as correlation function (ablation)SAC · CIFAR10AUC: 1.0 -> 1.0No clear effect
predicted-label-only access instead of probabilities (ablation)SAC · CIFAR10 Transfer-AAUC: 1.0 -> 0.0Hurt
predicted-label-only access instead of probabilities (ablation)SAC · CIFAR10 Transfer-LAUC: 1.0 -> 0.9No clear effect
SAC-m vs CAE fingerprint generationSAC-m · CIFAR10 timeseconds: 25,536.89s -> 4.45sImproved
EWE watermarking vs fingerprinting on source accuracyEWE · CIFAR10 source accuracyaccuracy: 87.3 -> 83.3(-4.0)Hurt

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