AI Paper Atlas

Research map / Adversarial robustness and security / Backdoor and poisoning

Poisoned Distillation: Injecting Backdoors into Distilled Datasets Without Raw Data Access

Ziyuan Yang, Ming Yan, Yi Zhang, Joey Tianyi Zhou

AAAI 2026 · Adversarial robustness and security › Backdoor and poisoning · approach: Detecting poisoned samples · 0 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
raise number of reconstructed archetypes m from 5 to 20 (ablation)Poisoned distillation attack · CIFAR-10 IPC1 (DC)benign accuracy (%): 25.79 -> 25.36No clear effect
raise number of reconstructed archetypes m from 5 to 20 (ablation)Poisoned distillation attack · CIFAR-10 IPC1 (DC)attack time (s): 41.5 -> 125.60Hurt
increase balance weight alpha toward backdoor loss (ablation)Poisoned distillation attack · CIFAR-10 IPC1 (DC)benign accuracy / ASRMixed
train attacker-side model as AlexNet instead of ConvNet (ablation)Poisoned distillation attack · CIFAR-10 IPC10 (DC)attack success rate (%): 100.00 -> 23.04Hurt
train attacker-side model as VGG11 instead of ConvNet (ablation)Poisoned distillation attack · CIFAR-10 IPC10 (DC)attack success rate (%): 100.00 -> 68.89Hurt
train attacker-side model as VGG11 instead of ConvNet (ablation)Poisoned distillation attack · CIFAR-10 IPC10 (MTT)attack success rate (%): 100.00 -> 77.21Hurt
increase user training epochs from 10 to 100Poisoned distillation attack · CIFAR-100 IPC1 (DC)attack success rate (%): 22.56 -> 99.83Improved
inject backdoor into benign distilled set (benign accuracy cost)Poisoned distillation attack · CIFAR-10 IPC10 (DC)benign accuracy (%): 42.63 -> 35.92Hurt
inject backdoor into benign distilled set (benign accuracy cost)Poisoned distillation attack · MNIST IPC1 (DC)benign accuracy (%): 88.81 -> 79.45Hurt
user trains ResNet18 instead of ConvNetPoisoned distillation attack · CIFAR-100 IPC1 (DC)attack success rate (%): 99.83 -> 0.03Hurt
user trains ResNet18 instead of ConvNetPoisoned distillation attack · CIFAR-10 IPC10 (DC)attack success rate (%): 100.00 -> 88.00Hurt
apply attack to MTT-distilled data with AlexNet user instead of ConvNetPoisoned distillation attack · CIFAR-10 IPC10 (MTT)attack success rate (%): 100.00 -> 54.31Hurt

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