AI Paper Atlas

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

Anti-Neuron Watermarking: Protecting Personal Data against Unauthorized Neural Networks

Zihang Zou, Boqing Gong, Liqiang Wang

ECCV 2022 · Adversarial robustness and security › Attacks on text and graphs · approach: Neural network watermarking · 12 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
add LCT watermark to user images (cosine similarity of inferred signature)ResNet50 on Tiny ImageNet · Tiny ImageNet (0.1% watermarked)cosine similarity: 0.045 -> 0.999Improved
add LCT watermarking vs dataset tracingTiny ImageNet · Tiny ImageNet (0.1% watermarked)cosine similarity after training: -0.005 -> 0.999Improved
add LCT watermarking to user dataResNet50 · Tiny ImageNet user datamemorization value estimate: 26.8% -> 34.8%Improved
reduce watermarked samples from 5 to 1 (ablation)LCT watermarking · Tiny ImageNet gradient searchsignature recovery: 60.99 matched -> 17.0 missedHurt
vary number of watermarked samples 10,000 to 5 (ablation)LCT watermarking · Tiny ImageNetmodel accuracy: 55.6 to 54.6No clear effect
vary the watermark signature (60 to 300) (ablation)LCT watermarking · Tiny ImageNetsignature recovery: inferred 59.5, 120.2, 178.8, 242.1, 306.1No clear effect
vary classifier architecture (Alex, VGG, ResNet, Wide Res, Dense) (ablation)LCT watermarking · Tiny ImageNetsignature recovery: inferred 56.9, 52.8, 59.5, 58.4, 58.8No clear effect
increase model capacity (ResNet18 to ResNet152) (ablation)LCT watermarking · Tiny ImageNetloss minimum sharpnessImproved
train with color jitter augmentation including hueLCT watermarking · Tiny ImageNetloss gap clean vs watermark: inferred 64.8 for signature 60Hurt
strong augmentation on CUB-200-Birds appendixLCT watermarking · CUB-200-Birdssignature recoveryHurt
more users with the same LCT (ablation)LCT watermarking · Tiny ImageNet 1,000 usersmatching accuracy: almost all correct at 20% users, drops significantly with same LCTHurt
user-specific random LCTs instead of the same LCT (ablation)LCT watermarking · Tiny ImageNet 1,000 usersmatching accuracy: above 80% with all users watermarkedImproved

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