Probing Adversarial Robustness of Protein Language Models: A Reproducible Case Study of ESM-2 Under Substitution-Based Attacks
Probing Adversarial Robustness of Protein Language Models: A Reproducible Case Study of ESM-2 Under Substitution-Based Attacks
Abstract
Description
Keywords
Deep Learning, ESM-2, Bayesian Uncertainty, Biosecurity, Pseudo-perplexity, Protein Language Models (PLMs), Machine Learning Security, Adversarial Robustness, Substitution-Based Attacks, Knowledge Distillation, Dual-Use Research, AI Governance
Fields of Science
0206 medical engineering, 02 engineering and technology
Citation
WoS Q
Scopus Q
Volume
24
Issue
2
Start Page
141
End Page
149