Enhancing Intrusion Detection Systems with Adaptive Learning Techniques
Naufal Zahir Rizqullah, Julius Alekhine, Dwika Lovitasari Yonia, Raden Mokhamad Racel Purnomo, Ary Mazharuddin Shiddiqi
Abstract
The importance of adaptive and deep learning in cybersecurity aims to advocate for various techniques to combat the constantly evolving cyber threats effectively. Those techniques utilize diverse attack types and network traffic data to mirror the real-world security environment accurately. This study proposes an intrusion detection technique using adaptive random forest (ARF) to enhance its performance. The ARF uses an ensemble learning method designed to handle data streams or changing data over time, thereby allowing the model to adapt and evolve. The experiment results indicate an increase in the accuracy of the ARF compared to its native Random Forest (RF) model, with accuracy rates of 97% and 99%, respectively. This improvement underscores ARF's effectiveness in enhancing the performance of intrusion detection systems.