| Autor: |
Kumar, A. Senthil, Nagalakshmi, T. J. |
| Předmět: |
|
| Zdroj: |
AIP Conference Proceedings; 2024, Vol. 2871 Issue 1, p1-8, 8p |
| Abstrakt: |
The goal of this research is to create a new intrusion detection system (IDS) by comparing two groups: one that uses Naive Bayes without feature removal and another that uses Naive Bayes with Fisher's Score Feature removal. The study divides Deep Machine Learning and Naive Bayes according to their use of intrusion detection systems. A significant level of 0.044 (p<0.05) was found for accuracy after an independent sample T test on both groups. To build an IDS, data was used from the NSL-KDD Dataset. We employed 38 samples in total, 19 from each group, for the analysis. For the statistical analysis, we relied on the SPSS program. Using G power, we approximated the sample size with an 80% pretest power. An increase in the percentage results in a statistically significant improvement (0.044) in the accuracy of IDS. Intruder detection systems (IDS) that use Naive Bayes with fisher's score feature elimination have an average accuracy of 0.7437, compared to 0.7432 when it is not used. using or not using fisher's score feature removal has no discernible effect on Naive Bayes performance. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
