Predicting the Traces of Adversaries using the Enhanced Data Encoding Method

Network intrusion detection research has turned into a hotspot for deep learning (DL). To significantly increase detection performance and accuracy, we provide an upgraded learning network-based intrusion detection model known as Enhanced Data Encoder (EDE). This approach can result in challenges, including the model converging to local optima or requiring extended training durations. To tackle these challenges, the EDE with the supervised learning-capable model. To solve the problems of poor classification performance, which often arises from EDE random kernel parameter initialization is proposed. Based on empirical tests on the BoT-IoT, IoT network dataset, MQTT, IoT-23, DS-2 and IoT-DS-1 datasets where the suggested EDE algorithm performs superior than present approaches in accuracy and other evaluation metrics.