SOTAVerified

Backdoor Attack

Backdoor attacks inject maliciously constructed data into a training set so that, at test time, the trained model misclassifies inputs patched with a backdoor trigger as an adversarially-desired target class.

Papers

Recently AddedMost HypedMost ActiveNeeds VerificationMost Verified

Showing 26–50 of 523 papers

TitleStatusHype
BadEdit: Backdooring large language models by model editingCode1
Mitigating Fine-tuning based Jailbreak Attack with Backdoor Enhanced Safety AlignmentCode1
Poisoned Forgery Face: Towards Backdoor Attacks on Face Forgery DetectionCode1
Model Supply Chain Poisoning: Backdooring Pre-trained Models via Embedding IndistinguishabilityCode1
Not All Prompts Are Secure: A Switchable Backdoor Attack Against Pre-trained Vision TransfomersCode1
FlowMur: A Stealthy and Practical Audio Backdoor Attack with Limited KnowledgeCode1
Universal Jailbreak Backdoors from Poisoned Human FeedbackCode1
BadCLIP: Dual-Embedding Guided Backdoor Attack on Multimodal Contrastive LearningCode1
Label Poisoning is All You NeedCode1
PoisonPrompt: Backdoor Attack on Prompt-based Large Language ModelsCode1
Composite Backdoor Attacks Against Large Language ModelsCode1
VDC: Versatile Data Cleanser based on Visual-Linguistic Inconsistency by Multimodal Large Language ModelsCode1
PatchBackdoor: Backdoor Attack against Deep Neural Networks without Model ModificationCode1
BAGM: A Backdoor Attack for Manipulating Text-to-Image Generative ModelsCode1
Backdooring Instruction-Tuned Large Language Models with Virtual Prompt InjectionCode1
You Can Backdoor Personalized Federated LearningCode1
Risk-optimized Outlier Removal for Robust 3D Point Cloud ClassificationCode1
Towards Stealthy Backdoor Attacks against Speech Recognition via Elements of SoundCode1
FedDefender: Backdoor Attack Defense in Federated LearningCode1
Bkd-FedGNN: A Benchmark for Classification Backdoor Attacks on Federated Graph Neural NetworkCode1
VillanDiffusion: A Unified Backdoor Attack Framework for Diffusion ModelsCode1
Backdoor Attack with Sparse and Invisible TriggerCode1
Text-to-Image Diffusion Models can be Easily Backdoored through Multimodal Data PoisoningCode1
UNICORN: A Unified Backdoor Trigger Inversion FrameworkCode1
Influencer Backdoor Attack on Semantic SegmentationCode1
Show:102550
← PrevPage 2 of 21Next →

No leaderboard results yet.