Can You Trust AI-assisted Network Automation? A DRL-based Approach to Mislead the Automation in SD-IPoEONs
Min Wang, Siqi Liu, Zuqing Zhu · 2020
We study the vulnerability of artificial intelligence assisted network automation (AIaNA), and design a deep reinforcement learning (DRL) model to mislead the AIaNA in software-defined IP over elastic optical networks (SD-IPoEONs) through crafting/injecting adversarial traffic samples.