SEM: Stacking Ensemble Meta-Learning for IOT Security Framework

Despite the popularity of the applicability of Internet of Things (IoT) devices in many applications, security and anomaly detection has always been a rising concern for any domain of research. In the current smart era, the increased use of IoT and related applications in almost all domains tune us for the alarming situations of security measures. Many such insecure threats such as the denial of service, malicious control, and operations can be the real harm for any IoT devices. Compared to the initial days of developments in IoT control, many advanced techniques based on machine learning are designed for effective control of such malicious activities. In this paper, a stacked ensemble meta-learning (SEM) model has been developed to enhance the performance of the base machine learning model used for anomaly detection in IoT devices. The proposed model learns from the prediction errors of the base classifiers to build a more accurate prediction model. The proposed SEM constructs a higher-level prediction model over the predictions of weak base classifiers.