Analysis of Accuracy in Anomaly Detection of Intrusion Detection System Using Naïve Bayes Algorithm Compared Over Gaussian Model
P.V. Pramila, M. Gayathri · ECS Transactions · 2022
Aim: To design a best intrusion model that monitors the malicious activities over the network. Intrusion Detection System (IDS) attempts to identify and notify the activities of users as normal (or) anomaly. Materials and Methods: Detection of anomalies for intrusion detection system is performed using Naive Bayes model and Gaussian model (sample size=20). Results: The model performance are validated using accuracy, F1 score, Precision metric. It is observed that innovative method Naive Bayes appears to be a better performing model with an accuracy of 95% when compared over Gaussian model (89%).The significance value obtained is 0.020 (p<0.05). Conclusion: Analysis of accuracy in anomaly detection of network intrusion detection systems is significantly better in Naive Bayes than Gaussian model. Empirical results of F1 score and precision are also observed higher in Naive Bayes model.