Pengembangan Model Machine Learning untuk Deteksi Serangan Siber

Nurdin Andi Baso Daeng Marewa (Universitas Kalbis)
Muhammad Adrinta Abdurrazzaq

KALBISIANA Jurnal Sains, Bisnis dan Teknologi
Vol. 12 No. 3 (2026): Kalbisiana

This study proposes a deep learning-based Intrusion Detection System (IDS) by combining Convolutional Neural Network (CNN) and Random Forest (RF) to detect network attacks. The CICIDS2018 dataset is used as training data to recognize various types of attacks such as DDoS and Brute Force. CNN acts as a feature extractor, while RF is used for classification. This system is implemented as a web application with an interactive interface for ease of use. Test results show that the hybrid CNN-RF model achieves high accuracy (91%) and is superior to both the single CNN model and CNN-XGBoost. This approach improves attack detection accuracy and provides an adaptive and efficient solution for network security.