A novel approach to constructing features and models for intrusion detection systems using SAE-ELM model
Davinder Paul Singh, Vinod Wamanrao Gangane, S. Praveen Kumar, Jyoti Kaushal, Neerav Nishant, Harshal Patil · 2025
Integrating intrusion detection (ID) into infrastructure security procedures is essential. It is essential for an intrusion detection system (IDS) to have accuracy, extensibility, and adaptability. Due to these demands and the complexity of modern network settings, a more structured and automated approach is needed to design intrusion detection systems (IDS). This is in contrast to the conventional methods, which rely purely on knowledge encoding and engineering. The three primary parts are training the model, feature selection, and preprocessing. The goal of data pre-processing is to make the raw data more usable and efficient for subsequent processing stages. Feature selection makes use of mutual information (MI) to assess any kind of random reliance between variables. For more accurate findings, we used SAE-ELM to train the model. The proposed method achieved a higher level of accuracy (96.30 percent) compared to its competitors, including SAE and ELM.