Taught Well Learned Ill: Towards Distillation-conditional Backdoor Attack

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chen, Yukun, Li, Boheng, Yuan, Yu, Qi, Leyi, Li, Yiming, Zhang, Tianwei, Qin, Zhan, Ren, Kui
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912612883628032
author Chen, Yukun
Li, Boheng
Yuan, Yu
Qi, Leyi
Li, Yiming
Zhang, Tianwei
Qin, Zhan
Ren, Kui
author_facet Chen, Yukun
Li, Boheng
Yuan, Yu
Qi, Leyi
Li, Yiming
Zhang, Tianwei
Qin, Zhan
Ren, Kui
contents Knowledge distillation (KD) is a vital technique for deploying deep neural networks (DNNs) on resource-constrained devices by transferring knowledge from large teacher models to lightweight student models. While teacher models from third-party platforms may undergo security verification (\eg, backdoor detection), we uncover a novel and critical threat: distillation-conditional backdoor attacks (DCBAs). DCBA injects dormant and undetectable backdoors into teacher models, which become activated in student models via the KD process, even with clean distillation datasets. While the direct extension of existing methods is ineffective for DCBA, we implement this attack by formulating it as a bilevel optimization problem and proposing a simple yet effective method (\ie, SCAR). Specifically, the inner optimization simulates the KD process by optimizing a surrogate student model, while the outer optimization leverages outputs from this surrogate to optimize the teacher model for implanting the conditional backdoor. Our SCAR addresses this complex optimization utilizing an implicit differentiation algorithm with a pre-optimized trigger injection function. Extensive experiments across diverse datasets, model architectures, and KD techniques validate the effectiveness of our SCAR and its resistance against existing backdoor detection, highlighting a significant yet previously overlooked vulnerability in the KD process. Our code is available at https://github.com/WhitolfChen/SCAR.
format Preprint
id arxiv_https___arxiv_org_abs_2509_23871
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Taught Well Learned Ill: Towards Distillation-conditional Backdoor Attack
Chen, Yukun
Li, Boheng
Yuan, Yu
Qi, Leyi
Li, Yiming
Zhang, Tianwei
Qin, Zhan
Ren, Kui
Cryptography and Security
Artificial Intelligence
Computer Vision and Pattern Recognition
Machine Learning
Knowledge distillation (KD) is a vital technique for deploying deep neural networks (DNNs) on resource-constrained devices by transferring knowledge from large teacher models to lightweight student models. While teacher models from third-party platforms may undergo security verification (\eg, backdoor detection), we uncover a novel and critical threat: distillation-conditional backdoor attacks (DCBAs). DCBA injects dormant and undetectable backdoors into teacher models, which become activated in student models via the KD process, even with clean distillation datasets. While the direct extension of existing methods is ineffective for DCBA, we implement this attack by formulating it as a bilevel optimization problem and proposing a simple yet effective method (\ie, SCAR). Specifically, the inner optimization simulates the KD process by optimizing a surrogate student model, while the outer optimization leverages outputs from this surrogate to optimize the teacher model for implanting the conditional backdoor. Our SCAR addresses this complex optimization utilizing an implicit differentiation algorithm with a pre-optimized trigger injection function. Extensive experiments across diverse datasets, model architectures, and KD techniques validate the effectiveness of our SCAR and its resistance against existing backdoor detection, highlighting a significant yet previously overlooked vulnerability in the KD process. Our code is available at https://github.com/WhitolfChen/SCAR.
title Taught Well Learned Ill: Towards Distillation-conditional Backdoor Attack
topic Cryptography and Security
Artificial Intelligence
Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2509.23871