Dataset Ownership Verification in Contrastive Pre-trained Models

2025-02-11

AI Summary

This work addresses dataset ownership verification for self-supervised contrastive learning models. We propose the first verifiable decision framework tailored to black-box backbone networks, enabling provenance attribution of specific pretraining datasets even in label-free settings. Our method leverages statistically significant disparities—quantified via p-values—in unary (single-instance) versus binary (positive/negative pair) instance relationships within the embedding space. The framework is model-agnostic and seamlessly integrates with mainstream self-supervised architectures, including SimCLR, BYOL, SimSiam, MoCo v3, and DINO. Extensive experiments demonstrate statistically significant verification (p < 0.05) across all evaluated models, consistently outperforming existing approaches in accuracy, robustness, and applicability. The implementation is publicly available.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Unsupervised & Self-Supervised Learning
• Data Mining & Knowledge Management: Representing, Reasoning, and Using Provenance, Trust

Application Category

• Security and Privacy: Data transparency and provenance
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

High-quality open-source datasets, which necessitate substantial efforts for curation, has become the primary catalyst for the swift progress of deep learning. Concurrently, protecting these datasets is paramount for the well-being of the data owner. Dataset ownership verification emerges as a crucial method in this domain, but existing approaches are often limited to supervised models and cannot be directly extended to increasingly popular unsupervised pre-trained models. In this work, we propose the first dataset ownership verification method tailored specifically for self-supervised pre-trained models by contrastive learning. Its primary objective is to ascertain whether a suspicious black-box backbone has been pre-trained on a specific unlabeled dataset, aiding dataset owners in upholding their rights. The proposed approach is motivated by our empirical insights that when models are trained with the target dataset, the unary and binary instance relationships within the embedding space exhibit significant variations compared to models trained without the target dataset. We validate the efficacy of this approach across multiple contrastive pre-trained models including SimCLR, BYOL, SimSiam, MOCO v3, and DINO. The results demonstrate that our method rejects the null hypothesis with a $p$-value markedly below $0.05$, surpassing all previous methodologies. Our code is available at https://github.com/xieyc99/DOV4CL.

Problem

• Verify dataset ownership in unsupervised models
• Tailored for contrastive learning pre-trained models
• Identify unlabeled dataset usage in black-box models

Innovation

• Contrastive learning verification method
• Unsupervised pre-trained models analysis
• Dataset ownership empirical validation