Anomaly Detection in Network Traffic Using LSTM with Attention Mechanism and Willow Catkin Optimization
Shashikant Verma, S. Prabakeran · 2024
A new anomaly detection method is presented in this study with the aim of improving the safety and reliability of IoT networks. The attention mechanism and Long Short-Term Memory (LSTM) networks are tuned using the new heuristic method known as Willow Catkin Optimization (WCO). LSTM networks excel in capturing the intricate temporal relationships seen in IoT traffic data, and the attention mechanism hones down on the most important aspects, enhancing the model's anomaly detection capabilities. Utilizing WCO, the hyperparameters of the LSTM network are fine-tuned to guarantee peak performance, significantly improving detection accuracy. Internet of Things (IoT) traffic is complex and ever-changing, which the suggested solution takes into account. Using a combination of feature prioritization and temporal pattern recognition, this method efficiently and reliably identifies possible security concerns in IoT contexts. Improving anomaly detection while minimizing false positives, the attention-enhanced LSTM model improved with WCO is particularly suitable for real-world IoT network security. More precise threat detection and improved network resilience are two ways this initiative improves upon previous efforts in the area of Internet of Things cybersecurity.