Anomaly detection in the internet of things by using machine learning techniques

Internet of Things (IoT) based systems, such as biometrics, robotic surgery, remote sensing, and other applications, deal with large amounts of data. Such applications face a high risk of security threats due to the large amounts of data in transit and at rest. We use Machine learning (ML) algorithms to enhance the security of the IoT system and prevent anomalous attacks with high accuracy. This manuscript proposes supervised and unsupervised learning methods for classification and clustering to identify and analyze anomaly attacks in IoT-based systems. We utilize Logistic Regression (LR), Random Forest (RF), and Naïve Bayes (NB) via an applied clustering technique using K-Means for categorization and prioritization. The proposed feature set is appropriate for use in an IoT-based system. Proper analysis of such attacks and improving anomaly detection in the IoT using predictive (ML) techniques are required to prevent them. Moreover, our solution shows better accuracy with lower overhead than existing solutions.