Enhancing security by monitoring the behavior of different classification algorithms
J Suriya Prakash, Thinley Tsering Lama, R. Soniya, V. Lakshmanan, S Kiran · 2025
Using Kyoto University’s Honeypots dataset, this study investigates how well different machine learning techniques identify and classify network intrusions. It compares the efficacy of various algorithms in detecting intrusions, including AdaBoost, CatBoost, KNN, LDA, logistic regression, LightGBM, LSVM, MLPC, Naive Bayes, quadratic discriminant analysis, XGBoost, decision tree, gradient boosting, KSVM, and random forest. To improve model accuracy, the study also uses preprocessing methods such feature scaling and imputation. The findings show that certain algorithms do better at spotting network intruders.