Answers are grounded in this paper and cite their sources.
Keylogger attack is a nefarious malware technique by which cybercriminals operate and thus advanced mitigation techniques against this attack are highly required. Machine learning as well as Deep learning algorithms with the ability to inspect data traffic and reveal abnormal patterns are highly influential in developing anomaly based intrusion detection systems against keylogger attacks. This study involves a benchmarking of performances with respect to four classical machine learning (ML) algorithms namely, Logistic Regression, K-nearest Neighbours, Random Forest and CatBoost as well as a Deep learning model (Artificial Neural Network). The models were used to fit the Keylogger_Detection dataset that originated from the Canadian Institute of Cybersecurity. In order to speed up computation, only ten percent of the data was employed in this study. Necessary data exploration, feature engineering and preprocessing such as Standard Scaling were carried out. The data was split into 70% train and 30% test portions, and a 10-fold cross validation was carried out on the training data to avoid overfitting. The models were evaluated on the bases of standard metrics such as Accuracy, Recall, Precision, F1-score and ROC-AUC. Random Forest was the optimal classifier based on pure classification Accuracy of 93.30% and balanced F1-Score of 91.60% at testing. Furthermore, with a test Precision score of 94.36%, it held the best compromise between avoiding false negatives and minimizing false alarms. However, the best model for high-security deployment (zero-tolerance for missed threats) was CatBoost, having the highest test Recall (approximately 92.5%).
No takes yet. Share an insight, caveat, or question.
Osa et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: