AI Paper Atlas

Research map / Adversarial robustness and security / Backdoor and poisoning

Finding Naturally Occurring Physical Backdoors in Image Datasets

Emily Wenger, Roma Bhattacharjee, Arjun Nitin Bhagoji, Josephine Passananti, Emilio Andere, Haitao Zheng, Ben Zhao

NeurIPS 2022 · Adversarial robustness and security › Backdoor and poisoning · approach: Trigger-based backdoor attacks · 2 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 the approximate MIS constraint when choosing the poisonable class subset (ablation)centrality-based natural backdoor dataset · Open Imagesclean accuracy: 58 ± 5% -> 72 ± 1%Improved
add the approximate MIS constraint when choosing the poisonable class subset (ablation)centrality-based natural backdoor dataset · Open Imagestrigger accuracy: 63 ± 8% -> 68 ± 3%Improved
natural backdoor dataset versus training without any backdoorOpen Images models · Open Imagesclean accuracy: 79 ± 2% -> 72 ± 1%Hurt
switch model architecture from VGG16 to ResNet (ablation)natural backdoor jeans dataset · Open Images jeans triggertrigger accuracy: 69 ± 5% -> 75 ± 4%Improved
switch model architecture from ResNet to Inception (ablation)natural backdoor jeans dataset · Open Images jeans triggertrigger accuracy: 75 ± 4% -> 61 ± 1%Hurt
vary model architecture (DenseNet, ResNet, VGG16, Inception) (ablation)natural backdoor jeans dataset · Open Images jeans triggerclean and trigger accuracyNo clear effect
increase poison injection rate (ablation)natural backdoor jeans dataset · Open Images jeans triggerclean and trigger accuracyImproved
use betweenness centrality instead of other centrality measures (ablation)trigger selection · Open Images and ImageNetclean/trigger accuracy consistencyImproved
use closeness centrality instead of other centrality measures (ablation)trigger selection · Open Imagesnumber of viable triggersHurt
remove spectral-flagged data and retrainnatural backdoor models · natural backdoor datasetsattack accuracy: 4%No clear effect
apply NeuralCleanse, Activation Clustering, Spectral, STRIP defensesnatural backdoor models · natural backdoor datasetsdefense performance: 16%, 9.7 ± 10.8%, 65.0 ± 4.3%, 4.0 ± 4.0%No clear effect
apply SentiNet defense (saliency maps)natural backdoor models · natural backdoor datasetsdefense performance: 56.9 ± 18.5%Improved

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