Machine Learning Approaches for Intrusion Detection System in Network Security


  • Vikash
  • Suruchi


Intrusion Detection System, Machine Learning, Datasets, feature selection.


IDS: An Intrusion Detection System refers to software used to monitor the entire network or the traffic to detect any type of malicious or abnormal activities. IDS plays an essential role in safeguarding computer networks against harmful activity. Traditional rule-based IDS such as firewalls are based on the method of data filtering, which is not capable of detecting all types of attacks & often encounters difficulties in adapting to the rapidly growing nature of the cyber threats. As a result, there’s been increasing interest in utilizing ML techniques to improve the functioning of the IDS. This ML-based detection system is used to detect the following 4 kinds of attacks in the network namely Pro, U2R (User to Root), R2L (Remote to Local), and Denial of Services. This study presents a novel ML strategy for IDS that makes use of a wide range of ML algorithms, like decision trees, k-Nearest Neighbors (KNN), SVMs, & naïve Bayes, etc. The proposed system extracts feature from the network traffic data and employs supervised ML techniques to classify instances as normal or malicious. Moreover, the system incorporates anomaly detection mechanisms to identify previously unseen attack patterns. We used the ‘NSL-KDD Dataset’ in that paper.

The outcomes of the experiments show how well the suggested method works in precisely identifying different kinds of intrusions while reducing false positives and adhering to performance metrics like F1 score, Precision, Recall, and Accuracy to evaluate each model's performance. Overall, this research helps make IDS better at keeping networks safe from cyber threats by using ML.


Download data is not yet available.




How to Cite

Vikash, & Suruchi. (2024). Machine Learning Approaches for Intrusion Detection System in Network Security. International Journal of Progressive Research in Science and Engineering, 5(05), 80–83. Retrieved from