How Much of My Dataset Did You Use? Quantitative Data Usage Inference in Machine Learning
Abstract
How much of my data was used to train a machine learning model? This is a critical question for data owners assessing the risk of unauthorized usage of their data to train models. However, previous work mistakenly treats this as a binary problem—inferring whether all-or-none or any-or-none of the data was used—which is fragile when faced with real, non-binary data usage risks. To address this, we propose a fine-grained analysis called Dataset Usage Cardinality Inference (DUCI), which estimates the exact proportion of data used. Our algorithm, leveraging debiased membership guesses, matches the performance of the optimal MLE approach (with a maximum error <0.1) but with significantly lower (e.g., $300 \times$ less) computational cost.
Cite
Text
Tong et al. "How Much of My Dataset Did You Use? Quantitative Data Usage Inference in Machine Learning." International Conference on Learning Representations, 2025.Markdown
[Tong et al. "How Much of My Dataset Did You Use? Quantitative Data Usage Inference in Machine Learning." International Conference on Learning Representations, 2025.](https://mlanthology.org/iclr/2025/tong2025iclr-much/)BibTeX
@inproceedings{tong2025iclr-much,
title = {{How Much of My Dataset Did You Use? Quantitative Data Usage Inference in Machine Learning}},
author = {Tong, Yao and Ye, Jiayuan and Zarifzadeh, Sajjad and Shokri, Reza},
booktitle = {International Conference on Learning Representations},
year = {2025},
url = {https://mlanthology.org/iclr/2025/tong2025iclr-much/}
}