Study and Detection of IOT Data Errors to avoid Malicious Activity

IoT anomalies are typically the result of malicious activity. For example, an attempted network intrusion could result in a point anomaly, while a device hacked could result in a mass anomaly. Due to the nature of attacks, some anomalies are represented by incomplete captured instances or imbalanced captured datasets. For example, a feature may be missing some values in rows, or it may contain both categorical and numeric values. After preprocessing, these datasets are good training sets for machine learning classifiers that detect anomalies. However, there are cases where preprocessing at the operation stage takes time, or where preprocessing is not possible due to the nature of the data. For example, features with an unknown number of categorical values. B. Strings are not converted to a finite number of binary features before training. In this scenario, basic machine learning methods such as support vector machines and decision trees either do not work or perform poorly for classification. Unlike Basic, Ensemble Learner manages these data instances efficiently and provides excellent anomaly detection rates.