Image de couverture de livre

Scaling Exposes the Trigger: Input-Level Backdoor Detection in Text-to-Image Diffusion Models via Cross-Attention Scaling

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Li, Zida, Li, Jun, Sha, Yuzhe, Li, Ziqiang, Xiong, Lizhi, Fu, Zhangjie
Format:	Preprint
Publié:	2026
Sujets:	Cryptography and Security Computer Vision and Pattern Recognition
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Beyond Text Prompts: Precise Concept Erasure through Text-Image Collaboration
par: Li, Jun, et autres
Publié: (2026)

Is Artificial Intelligence Generated Image Detection a Solved Problem?
par: Li, Ziqiang, et autres
Publié: (2025)

Large Language Models are Good Attackers: Efficient and Stealthy Textual Backdoor Attacks
par: Li, Ziqiang, et autres
Publié: (2024)

IBD-PSC: Input-level Backdoor Detection via Parameter-oriented Scaling Consistency
par: Hou, Linshan, et autres
Publié: (2024)

Real is not True: Backdoor Attacks Against Deepfake Detection
par: Sun, Hong, et autres
Publié: (2024)

Efficient Input-level Backdoor Defense on Text-to-Image Synthesis via Neuron Activation Variation
par: Zhai, Shengfang, et autres
Publié: (2025)

Isolate Trigger: Detecting and Eliminating Adaptive Backdoor Attacks
par: Sun, Chengrui, et autres
Publié: (2025)

BackdoorDM: A Comprehensive Benchmark for Backdoor Learning on Diffusion Model
par: Lin, Weilin, et autres
Publié: (2025)

Cross-Paradigm Graph Backdoor Attacks with Promptable Subgraph Triggers
par: Liu, Dongyi, et autres
Publié: (2025)

Meme Trojan: Backdoor Attacks Against Hateful Meme Detection via Cross-Modal Triggers
par: Wang, Ruofei, et autres
Publié: (2024)

A Proxy Attack-Free Strategy for Practically Improving the Poisoning Efficiency in Backdoor Attacks
par: Li, Ziqiang, et autres
Publié: (2023)

Exposing Vulnerabilities in RL: A Novel Stealthy Backdoor Attack through Reward Poisoning
par: Zhang, Bokang, et autres
Publié: (2025)

Practical, Generalizable and Robust Backdoor Attacks on Text-to-Image Diffusion Models
par: Dai, Haoran, et autres
Publié: (2025)

Backdoor Sentinel: Detecting and Detoxifying Backdoors in Diffusion Models via Temporal Noise Consistency
par: Wang, Bingzheng, et autres
Publié: (2026)

CL-Attack: Textual Backdoor Attacks via Cross-Lingual Triggers
par: Zheng, Jingyi, et autres
Publié: (2024)

Awakening the Hydra: Stabilizing Multi-Concept Backdoor Injection in Text-to-Image Diffusion Models
par: Wang, Kai, et autres
Publié: (2026)

Combinational Backdoor Attack against Customized Text-to-Image Models
par: Jiang, Wenbo, et autres
Publié: (2024)

Backdoor Token Unlearning: Exposing and Defending Backdoors in Pretrained Language Models
par: Jiang, Peihai, et autres
Publié: (2025)

Cross-Modal Backdoors in Multimodal Large Language Models
par: Wang, Runhe, et autres
Publié: (2026)

BackdoorIndicator: Leveraging OOD Data for Proactive Backdoor Detection in Federated Learning
par: Li, Songze, et autres
Publié: (2024)

Shortcuts Everywhere and Nowhere: Exploring Multi-Trigger Backdoor Attacks
par: Li, Yige, et autres
Publié: (2024)

Backdoor Contrastive Learning via Bi-level Trigger Optimization
par: Sun, Weiyu, et autres
Publié: (2024)

ProP: Efficient Backdoor Detection via Propagation Perturbation for Overparametrized Models
par: Ren, Tao, et autres
Publié: (2024)

UFID: A Unified Framework for Input-level Backdoor Detection on Diffusion Models
par: Guan, Zihan, et autres
Publié: (2024)

STEP: Detecting Audio Backdoor Attacks via Stability-based Trigger Exposure Profiling
par: Wang, Kun, et autres
Publié: (2026)

Attack by Yourself: Effective and Unnoticeable Multi-Category Graph Backdoor Attacks with Subgraph Triggers Pool
par: Li, Jiangtong, et autres
Publié: (2024)

AutoBackdoor: Automating Backdoor Attacks via LLM Agents
par: Li, Yige, et autres
Publié: (2025)

Robustness of Watermarking on Text-to-Image Diffusion Models
par: Wu, Xiaodong, et autres
Publié: (2024)

Hardware-Triggered Backdoors
par: Möller, Jonas, et autres
Publié: (2026)

Fusing Pruned and Backdoored Models: Optimal Transport-based Data-free Backdoor Mitigation
par: Lin, Weilin, et autres
Publié: (2024)

Semantic-level Backdoor Attack against Text-to-Image Diffusion Models
par: Chen, Tianxin, et autres
Publié: (2026)

REDEditing: Relationship-Driven Precise Backdoor Poisoning on Text-to-Image Diffusion Models
par: Guo, Chongye, et autres
Publié: (2025)

Pay Attention to the Triggers: Constructing Backdoors That Survive Distillation
par: De Muri, Giovanni, et autres
Publié: (2025)

M-to-N Backdoor Paradigm: A Multi-Trigger and Multi-Target Attack to Deep Learning Models
par: Hou, Linshan, et autres
Publié: (2022)

Removing the Trigger, Not the Backdoor: Alternative Triggers and Latent Backdoors
par: Abad, Gorka, et autres
Publié: (2026)

Discovering Universal Semantic Triggers for Text-to-Image Synthesis
par: Zhai, Shengfang, et autres
Publié: (2024)

The Trigger in the Haystack: Extracting and Reconstructing LLM Backdoor Triggers
par: Bullwinkel, Blake, et autres
Publié: (2026)

Re-Triggering Safeguards within LLMs for Jailbreak Detection
par: Lin, Zheng, et autres
Publié: (2026)

BaThe: Defense against the Jailbreak Attack in Multimodal Large Language Models by Treating Harmful Instruction as Backdoor Trigger
par: Chen, Yulin, et autres
Publié: (2024)

When Forgetting Triggers Backdoors: A Clean Unlearning Attack
par: Arazzi, Marco, et autres
Publié: (2025)