Derui Wang (CSIRO's Data61), Minhui Xue (CSIRO's Data61), Bo Li (The University of Chicago), Seyit Camtepe (CSIRO's Data61), Liming Zhu (CSIRO's Data61)

The exploitation of publicly accessible data has led to escalating concerns regarding data privacy and intellectual property (IP) breaches in the age of artificial intelligence. To safeguard both data privacy and IP-related domain knowledge, efforts have been undertaken to render shared data unlearnable for unauthorized models in the wild. Existing methods apply empirically optimized perturbations to the data in the hope of disrupting the correlation between the inputs and the corresponding labels such that the data samples are converted into Unlearnable Examples (UEs). Nevertheless, the absence of mechanisms to verify the robustness of UEs against uncertainty in unauthorized models and their training procedures engenders several under-explored challenges. First, it is hard to quantify the unlearnability of UEs against unauthorized adversaries from different runs of training, leaving the soundness of the defense in obscurity. Particularly, as a prevailing evaluation metric, empirical test accuracy faces generalization errors and may not plausibly represent the quality of UEs. This also leaves room for attackers, as there is no rigid guarantee of the maximal test accuracy achievable by attackers. Furthermore, we find that a simple recovery attack can restore the clean-task performance of the classifiers trained on UEs by slightly perturbing the learned weights. To mitigate the aforementioned problems, in this paper, we propose a mechanism for certifying the so-called $(q, eta)$-Learnability of an unlearnable dataset via parametric smoothing. A lower certified $(q, eta)$-Learnability indicates a more robust and effective protection over the dataset. Concretely, we 1) improve the tightness of certified $(q, eta)$-Learnability and 2) design Provably Unlearnable Examples (PUEs) which have reduced $(q, eta)$-Learnability. According to experimental results, PUEs demonstrate both decreased certified $(q, eta)$-Learnability and enhanced empirical robustness compared to existing UEs. Compared to the competitors on classifiers with uncertainty in parameters, PUEs reduce at most $18.9%$ of certified $(q, eta)$-Learnability on ImageNet and $54.4%$ of the empirical test accuracy score on CIFAR-100. Our source code is available at https://github.com/NeuralSec/certified-data-learnability.

View More Papers

Towards Anonymous Chatbots with (Un)Trustworthy Browser Proxies

Dzung Pham, Jade Sheffey, Chau Minh Pham, and Amir Houmansadr (University of Massachusetts Amherst)

Read More

Hidden and Lost Control: on Security Design Risks in...

Haoqiang Wang (Institute of Information Engineering, Chinese Academy of Sciences; School of Cyber Security, University of Chinese Academy of Sciences; Indiana University Bloomington), Yiwei Fang (Institute of Information Engineering, Chinese Academy of Sciences; School of Cyber Security, University of Chinese Academy of Sciences; Indiana University Bloomington), Yichen Liu (Indiana University Bloomington), Ze Jin (Institute of…

Read More

JBomAudit: Assessing the Landscape, Compliance, and Security Implications of...

Yue Xiao (IBM Research), Dhilung Kirat (IBM Research), Douglas Lee Schales (IBM Research), Jiyong Jang (IBM Research), Luyi Xing (Indiana University Bloomington), Xiaojing Liao (Indiana University)

Read More

Moneta: Ex-Vivo GPU Driver Fuzzing by Recalling In-Vivo Execution...

Joonkyo Jung (Department of Computer Science, Yonsei University), Jisoo Jang (Department of Computer Science, Yonsei University), Yongwan Jo (Department of Computer Science, Yonsei University), Jonas Vinck (DistriNet, KU Leuven), Alexios Voulimeneas (CYS, TU Delft), Stijn Volckaert (DistriNet, KU Leuven), Dokyung Song (Department of Computer Science, Yonsei University)

Read More