Hidden Trigger Backdoor Attacks¶
Saha, Subramanya, & Pirsiavash. (2020). Hidden Trigger Backdoor Attacks. Proceedings of the AAAI Conference on Artificial Intelligence.
Cited by¶
1 citation across 1 artifact.
Each citation links to the sentence it supports in the citing article.
Domain-specific¶
- Data Poisoning Attack
- Clean-label backdoor attacks, which build on the triggerless clean-label poisoning of Ali Shafahi and colleagues and were developed in the label-consistent attacks of Turner, Tsipras, and Madry and the Hidden Trigger Backdoor framework of Aniruddha Saha and colleagues, plant training samples that appear correctly labelled and are benign to human inspection but carry an imperceptible trigger pattern — spatial, spectral, or stylistic — so that at inference time inputs carrying the trigger are misclassified to an attacker-chosen target while the model behaves normally on all other inputs
This sourceThe Hidden Trigger Backdoor attack, in which correctly labelled, natural-looking poisons hide a patch trigger that at test time flips source-class images to an attacker-chosen target while clean accuracy is preserved, building on Shafahi and colleagues' triggerless clean-label poisoning.
Supported in partVerified against the source
- Clean-label backdoor attacks, which build on the triggerless clean-label poisoning of Ali Shafahi and colleagues and were developed in the label-consistent attacks of Turner, Tsipras, and Madry and the Hidden Trigger Backdoor framework of Aniruddha Saha and colleagues, plant training samples that appear correctly labelled and are benign to human inspection but carry an imperceptible trigger pattern — spatial, spectral, or stylistic — so that at inference time inputs carrying the trigger are misclassified to an attacker-chosen target while the model behaves normally on all other inputs
Verification¶
Does it exist? Not checked yet. This work's DOI is recorded above but has not been resolved against an external catalogue, so nothing here confirms the work exists.
Does it back the claim? Read against the text for 1 of 1 citation: 1 supported in part. Each verdict is shown under its citation below, with what in the work backs the sentence.
Support is checked per citation rather than per work — the same source can be cited soundly in one article and wrongly in another. Per-citation recording began recently, so a citation with no recorded check is a gap in the record rather than evidence it went unchecked.
See how references were verified.
Registry ID ref:0e0d58afe4df · see in the full table