The widespread adoption of 5G networks and the emergence of advanced technologies have significantly fueled the rapid growth of the Internet of Things (IoT). IoT infrastructure is now integral across all industries and domains, increasing device communications, protocol complexity, and diversity. However, this has also given rise to more sophisticated attack methods that threaten system security. Unlike traditional network technologies, IoT systems’ attack surface and vulnerability space are exponentially larger, necessitating specialized security measures. This study proposes an innovative solution for intrusion detection at the edge system level. The Cerebellar Model Articulation Controller (CMAC) neural network and Random Forest and Multi-Layer Perceptron (MLP)-based hybrid feature selection techniques are used in our method. This methodology enables explicit and interpretable feature selection while incorporating the CMAC model to automate attack detection, reduce false positives, and enhance accuracy, aiming at detecting attacks at edge computing devices.