Poison-then-Hide: Finetuning-Activated Backdoor Attack on Pretrained Vision Encoders
cs.LG, cs.CR, cs.CV
Submitted: 2026-11-11
Updated: 2026-11-11
Code: https://github.com/QixuanJin99/poison-then-hide
Terminology
Sources
- Harmful Fine-tuning Attacks and Defenses for Large Language Models: A Survey
- Universal Vulnerabilities in Large Language Models: Backdoor Attacks for In-context Learning
- Backdoor Attacks on Pre-trained Models by Layerwise Weight Poisoning
- Watch your steps: Dormant Adversarial Behaviors that Activate upon LLM Finetuning
- DINOv2: Learning Robust Visual Features without Supervision
- BadNets: Identifying Vulnerabilities in the Machine Learning Model Supply Chain
- Witches' Brew: Industrial Scale Data Poisoning via Gradient Matching
- WaNet -- Imperceptible Warping-based Backdoor Attack
- Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoning
- Fine-Tuning Is All You Need to Mitigate Backdoor Attacks
- Backdoor Pre-trained Models Can Transfer to All
- Reproducible scaling laws for contrastive language-image learning
- Neural Attention Distillation: Erasing Backdoor Triggers from Deep Neural Networks
Related papers
- A Physics-Regulated Neural Framework for Learning 3D Grain Growth Dynamics
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis