Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2507.22813 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909712433283072 |
|---|---|
| author | Mirzaei, Hossein Taghavi, Zeinab Rezaee, Sepehr Hadi, Masoud Madadi, Moein Mathis, Mackenzie W. |
| author_facet | Mirzaei, Hossein Taghavi, Zeinab Rezaee, Sepehr Hadi, Masoud Madadi, Moein Mathis, Mackenzie W. |
| contents | Deep neural networks have demonstrated remarkable success across numerous tasks, yet they remain vulnerable to Trojan (backdoor) attacks, raising serious concerns about their safety in real-world mission-critical applications. A common countermeasure is trigger inversion -- reconstructing malicious "shortcut" patterns (triggers) inserted by an adversary during training. Current trigger-inversion methods typically search the full pixel space under specific assumptions but offer no assurances that the estimated trigger is more than an adversarial perturbation that flips the model output. Here, we propose a data-free, zero-shot trigger-inversion strategy that restricts the search space while avoiding strong assumptions on trigger appearance. Specifically, we incorporate a diffusion-based generator guided by the target classifier; through iterative generation, we produce candidate triggers that align with the internal representations the model relies on for malicious behavior. Empirical evaluations, both quantitative and qualitative, show that our approach reconstructs triggers that effectively distinguish clean versus Trojaned models. DISTIL surpasses alternative methods by high margins, achieving up to 7.1% higher accuracy on the BackdoorBench dataset and a 9.4% improvement on trojaned object detection model scanning, offering a promising new direction for reliable backdoor defense without reliance on extensive data or strong prior assumptions about triggers. The code is available at https://github.com/AdaptiveMotorControlLab/DISTIL. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_22813 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | DISTIL: Data-Free Inversion of Suspicious Trojan Inputs via Latent Diffusion Mirzaei, Hossein Taghavi, Zeinab Rezaee, Sepehr Hadi, Masoud Madadi, Moein Mathis, Mackenzie W. Computer Vision and Pattern Recognition Deep neural networks have demonstrated remarkable success across numerous tasks, yet they remain vulnerable to Trojan (backdoor) attacks, raising serious concerns about their safety in real-world mission-critical applications. A common countermeasure is trigger inversion -- reconstructing malicious "shortcut" patterns (triggers) inserted by an adversary during training. Current trigger-inversion methods typically search the full pixel space under specific assumptions but offer no assurances that the estimated trigger is more than an adversarial perturbation that flips the model output. Here, we propose a data-free, zero-shot trigger-inversion strategy that restricts the search space while avoiding strong assumptions on trigger appearance. Specifically, we incorporate a diffusion-based generator guided by the target classifier; through iterative generation, we produce candidate triggers that align with the internal representations the model relies on for malicious behavior. Empirical evaluations, both quantitative and qualitative, show that our approach reconstructs triggers that effectively distinguish clean versus Trojaned models. DISTIL surpasses alternative methods by high margins, achieving up to 7.1% higher accuracy on the BackdoorBench dataset and a 9.4% improvement on trojaned object detection model scanning, offering a promising new direction for reliable backdoor defense without reliance on extensive data or strong prior assumptions about triggers. The code is available at https://github.com/AdaptiveMotorControlLab/DISTIL. |
| title | DISTIL: Data-Free Inversion of Suspicious Trojan Inputs via Latent Diffusion |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2507.22813 |