Exploring the Efficacy of Deep Learning Techniques for IoT Intrusion Detection: a Study of Customized Vision Transformers

In today's hyperconnected world, the security of computer networks is paramount. This is especially true in the Internet of Things (IoT) age, where billions of heterogeneous devices are connected to the internet. This creates a vast and complex network that is increasingly vulnerable to cyberattacks. To mitigate these risks, it is essential to have an effective intrusion detection system (IDS). However, traditional IDSs do not have the capabilities to be up to date and to catch up with the dynamic nature of IoT threats. One of the new deep learning models that has shown promise for IDS in IoT environments is Vision Transformers (ViTs). This paper explores the impact of using vision transformers to detect intrusions in IoT environments. The results show promising applications of the vision transformers in the IDS domain.