High Speed Intrusion Detection Using Lexicographic Method in Compared Over Artificial Neural Networks with Better Accuracy

The primary goal of this research is to evaluate the effectiveness of the Lexicographic Method for intrusion detection and compare its performance with Artificial Neural Networks (ANNs). The study involved a sample size of 10 for each method, and we tested these methods multiple times to measure their accuracy. We used a pretest power analysis with an 80 % power level and a 95 % confidence interval to ensure our results were reliable. In the Lexicographic Method, the approach involves using multiple decision trees. Each tree generates its own classification, and the final decision is made based on which classification receives the most votes from these trees. Our findings indicate that there is a statistically insignificant difference between the Lexicographic Method and ANNs, with a p-value of 0.042, which is below the threshold of 0.05. However, the Lexicographic Method achieved an accuracy of 86.69 %, which is notably higher than the 84.65 % accuracy achieved by ANNs. This suggests that the Lexicographic Method is a more effective and straightforward approach for building fast and accurate machine learning models for intrusion detection.