I2I Backdoor: Backdoor Attacks against Image-to-Image Tasks
cs.CV, cs.AI, cs.CR
Submitted: 2024-07-15
Updated: 2026-11-18
Terminology
Sources
- Practical Blind Image Denoising via Swin-Conv-UNet and Data Synthesis
- Deep Image Destruction: Vulnerability of Deep Image-to-Image Models against Adversarial Attacks
- Towards Adversarially Robust Deep Image Denoising
- BAAAN: Backdoor Attacks Against Autoencoder and GAN-Based Machine Learning Models
- Rickrolling the Artist: Injecting Backdoors into Text Encoders for Text-to-Image Synthesis
- Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoning
- Distilling the Knowledge in a Neural Network
- Compression-Resistant Backdoor Attack against Deep Neural Networks
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs