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

TL;DR

DF-LoGiT is a novel data-free backdoor attack method on Vision Transformers that directly edits model weights to embed stealthy, logic-gated triggers, achieving high success rates without additional data or model components.

Contribution

It introduces the first truly data-free backdoor attack on ViTs using weight editing and exploits the multi-head architecture for logic-gated triggers.

Findings

Achieves near-100% attack success rate.

Maintains high benign accuracy.

Remains robust against defenses.

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.