Evaluasi Performa Model Ensemble Learning dalam Deteksi Serangan Jaringan Internet of Things pada Dataset CIC-BCCC-IOT-HCRL-2019

Abstrak

Abstract - The rapid growth of Internet of Things (IoT) devices has led to increased network complexity and higher risks of cybersecurity attacks. This research assesses the effectiveness of five Ensemble Learning algorithms—Random Forest, AdaBoost, CatBoost, XGBoost, and LightGBM—in Intrusion Detection Systems (IDS) for IoT networks, using the CIC-BCCC-IoT-HCRL-2019 dataset. The approach includes data preprocessing, implementing two normalization methods (MinMaxScaler and Normalizer), and evaluating models via 5-Fold Cross-Validation alongside an 80:20 train-test split. Results demonstrate that boosting algorithms like XGBoost, CatBoost, and LightGBM consistently outperform the bagging-based Random Forest. Specifically, XGBoost paired with MinMaxScaler reached the highest accuracy at 0.9980, whereas LightGBM combined with MinMaxScaler achieved the quickest training time of 2.54 seconds. This study highlights that integrating boosting algorithms with MinMaxScaler normalization significantly improves both the accuracy and speed of IoT intrusion detection.

Keywords - Internet of Things, Intrusion Detection, Machine Learning, Boosting, Data Normalization.