AI Paper Atlas

Research map / Adversarial robustness and security / Backdoor and poisoning

Handcrafted Backdoors in Deep Neural Networks

Sanghyun Hong, Nicholas Carlini, Alexey Kurakin

NeurIPS 2022 · Adversarial robustness and security › Backdoor and poisoning · approach: Trigger-based backdoor attacks · 19 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
handcrafted backdoor instead of poisoningBackdoor attack on pre-trained models · SVHN CNN, Checkerboardattack success rate (%): 96 -> 100Improved
handcrafted backdoor instead of poisoningBackdoor attack · CIFAR10 CNN, Checkerboardattack success rate (%): 98 -> 97Hurt
handcrafted backdoor instead of poisoning with few samplesBackdoor attack · CIFAR10attack success rate (%): 100% success with only 50 samplesImproved
handcrafted backdoor instead of poisoning under fine-tuningBackdoor attack · Fine-tuning defenseattack success rate (%): 82 -> 88 (CIFAR10 Checkerboard)Improved
handcrafted backdoor instead of poisoning under fine-pruningBackdoor attack · Fine-pruning on CIFAR10 and SVHNattack success rate (%): 96 -> 82 (CIFAR10 Random); 9-27% higher on SVHNMixed
add random noise or parameter clipping defensesHandcrafted backdoor · Defense at 5% accuracy dropattack success rate: over 98% (noise), over 86% (clipping)No clear effect

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