When Priors Backfire: On the Vulnerability of Unlearnable Examples to Pretraining

TL;DR

This paper reveals that unlearnable examples can be bypassed when models are pretrained, and introduces BAIT, a bi-level optimization method that enforces mislabel-perturbation binding to preserve data unlearnability.

Contribution

The paper uncovers a vulnerability of unlearnable examples with pretrained models and proposes BAIT, a novel method to maintain unlearnability by disrupting semantic learning.

Findings

Pretraining priors can nullify unlearnability by capturing genuine features.

BAIT effectively prevents models from learning true semantics from protected data.

Experiments show BAIT maintains unlearnability across benchmarks and backbones.

Abstract

Unlearnable Examples (UEs) serve as a data protection strategy that generates imperceptible perturbations to mislead models into learning spurious correlations instead of underlying semantics. In this paper, we uncover a fundamental vulnerability of UEs that emerges when learning starts from a pretrained model. Crucially, our empirical analysis shows that even when data are protected by carefully crafted perturbations, pretraining priors still furnish rich semantic representations that allow the model to circumvent the shortcuts introduced by UEs and capture genuine features, thereby nullifying unlearnability. To address this, we propose BAIT (Binding Artificial perturbations to Incorrect Targets), a novel bi-level optimization formulation. Specifically, the inner level aims at associating the perturbed samples with real labels to simulate standard data-label alignment, while the outer level actively disrupts this alignment by enforcing a mislabel-perturbation binding that maps samples to designated incorrect targets. This mechanism effectively overrides the semantic guidance of priors, forcing the model to rely on the injected perturbations and consequently preventing the acquisition of true semantics. Extensive experiments on standard benchmarks and multiple pretrained backbones demonstrate that BAIT effectively mitigates the influence of pretraining priors and maintains data unlearnability.