Securely Fine-tuning Pre-trained Encoders Against Adversarial Examples

arXiv:2403.10801v1 Announce Type: new
Abstract: With the evolution of self-supervised learning, the pre-training paradigm has emerged as a predominant solution within the deep learning landscape. Model providers furnish pre-trained encoders designed to function as versatile feature extractors, enabling downstream users to harness the benefits of expansive models with minimal effort through fine-tuning. Nevertheless, recent works have exposed a vulnerability in pre-trained encoders, highlighting their susceptibility to downstream-agnostic adversarial examples (DAEs) meticulously crafted by attackers. The lingering question pertains to the …

abstract adversarial adversarial examples arxiv benefits cs.cv deep learning enabling evolution examples feature fine-tuning function harness landscape paradigm pre-training self-supervised learning solution supervised learning through training type