DF-LoGiT: Data-Free Logic-Gated Backdoor Attacks in Vision Transformers

2026-02-03

AI Summary

This work addresses the risk of supply chain backdoor attacks against Vision Transformers (ViTs) in third-party model repositories. While existing approaches typically rely on fine-tuning with poisoned or synthetic data, this paper proposes the first truly data-free backdoor attack method. By directly manipulating model weights and exploiting the inherent multi-head attention architecture of ViTs, the method constructs logic-gated composite triggers without introducing additional modules or requiring fine-tuning. The approach achieves near-perfect attack success rates across multiple benchmarks while exerting negligible impact on the model’s performance on clean tasks. Furthermore, it demonstrates strong robustness against a variety of general and ViT-specific defense mechanisms, significantly enhancing both the practicality and stealth of backdoor attacks in real-world deployment scenarios.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Safety and Robustness

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

The widespread adoption of Vision Transformers (ViTs) elevates supply-chain risk on third-party model hubs, where an adversary can implant backdoors into released checkpoints. Existing ViT backdoor attacks largely rely on poisoned-data training, while prior data-free attempts typically require synthetic-data fine-tuning or extra model components. This paper introduces Data-Free Logic-Gated Backdoor Attacks (DF-LoGiT), a truly data-free backdoor attack on ViTs via direct weight editing. DF-LoGiT exploits ViT's native multi-head architecture to realize a logic-gated compositional trigger, enabling a stealthy and effective backdoor. We validate its effectiveness through theoretical analysis and extensive experiments, showing that DF-LoGiT achieves near-100% attack success with negligible degradation in benign accuracy and remains robust against representative classical and ViT-specific defenses.

Problem

• backdoor attack
• Vision Transformers
• data-free
• supply-chain security
• weight editing

Innovation

• Data-Free Backdoor
• Vision Transformers
• Weight Editing
• Logic-Gated Trigger
• Supply-Chain Attack