Yugeng Liu (CISPA Helmholtz Center for Information Security), Zheng Li (CISPA Helmholtz Center for Information Security), Michael Backes (CISPA Helmholtz Center for Information Security), Yun Shen (Netapp), Yang Zhang (CISPA Helmholtz Center for Information Security)

Dataset distillation has emerged as a prominent technique to improve data efficiency when training machine learning models. It encapsulates the knowledge from a large dataset into a smaller synthetic dataset. A model trained on this smaller distilled dataset can attain comparable performance to a model trained on the original training dataset. However, the existing dataset distillation techniques mainly aim at achieving the best trade-off between resource usage efficiency and model utility. The security risks stemming from them have not been explored. This study performs the first backdoor attack against the models trained on the data distilled by dataset distillation models in the image domain. Concretely, we inject triggers into the synthetic data during the distillation procedure rather than during the model training stage, where all previous attacks are performed. We propose two types of backdoor attacks, namely NAIVEATTACK and DOORPING. NAIVEATTACK simply adds triggers to the raw data at the initial distillation phase, while DOORPING iteratively updates the triggers during the entire distillation procedure. We conduct extensive evaluations on multiple datasets, architectures, and dataset distillation techniques. Empirical evaluation shows that NAIVEATTACK achieves decent attack success rate (ASR) scores in some cases, while DOORPING reaches higher ASR scores (close to 1.0) in all cases. Furthermore, we conduct a comprehensive ablation study to analyze the factors that may affect the attack performance. Finally, we evaluate multiple defense mechanisms against our backdoor attacks and show that our attacks can practically circumvent these defense mechanisms.

View More Papers

Lightning Community Shout-Outs to:

(1) Jonathan Petit, Secure ML Performance Benchmark (Qualcomm) (2) David Balenson, The Road to Future Automotive Research Datasets: PIVOT Project and Community Workshop (USC Information Sciences Institute) (3) Jeremy Daily, CyberX Challenge Events (Colorado State University) (4) Mert D. Pesé, DETROIT: Data Collection, Translation and Sharing for Rapid Vehicular App Development (Clemson University) (5) Ning…

Read More

Cooperative Perception for Safe Control of Autonomous Vehicles under...

Hongchao Zhang (Washington University in St. Louis), Zhouchi Li (Worcester Polytechnic Institute), Shiyu Cheng (Washington University in St. Louis), Andrew Clark (Washington University in St. Louis)

Read More

CHKPLUG: Checking GDPR Compliance of WordPress Plugins via Cross-language...

Faysal Hossain Shezan (University of Virginia), Zihao Su (University of Virginia), Mingqing Kang (Johns Hopkins University), Nicholas Phair (University of Virginia), Patrick William Thomas (University of Virginia), Michelangelo van Dam (in2it), Yinzhi Cao (Johns Hopkins University), Yuan Tian (UCLA)

Read More

Focusing on Pinocchio's Nose: A Gradients Scrutinizer to Thwart...

Jiayun Fu (Huazhong University of Science and Technology), Xiaojing Ma (Huazhong University of Science and Technology), Bin B. Zhu (Microsoft Research Asia), Pingyi Hu (Huazhong University of Science and Technology), Ruixin Zhao (Huazhong University of Science and Technology), Yaru Jia (Huazhong University of Science and Technology), Peng Xu (Huazhong University of Science and Technology), Hai…

Read More