Refining Intrusion Detection Capabilities Through Combined Algorithmic Classification Techniques
Chandini Lekkalapudi, Niranjana Holla V P, S Jagannathan, C Vasanthakumar, Suriya Prakash J · 2024
This research analyses the numerous ways in which IntrusionDetection Systems (IDS) can be enhanced via Machine Learning (ML) methodologies. Different classification techniques have been evaluated with the Kyoto20151208 dataset, such as logistic regression and random forest. Through evaluations of performance against various attack types, our aim is to enable cybersecurity experts to select the most effective machine learning model for their unique needs. In the end, this will increase the effectiveness of IDS overall. These algorithms are important because they are extremely effective at recognising trends in big datasets of network traffic. This correlates to higher processing speeds and provides them the ability to detect anomalies more effectively, rendering them fit for real-time intrusion detection applications. We also evaluate these algorithms' performances against various types of intrusions and network systems, extending beyond basic detection. With this indepth understanding, cybersecurity experts may select the best machine learning models for their specific network environment and safety needs. The study gives them the ability to improve their IDS the effectiveness and improve the overall cybersecurity posture of their organisation by giving them exposure to this actionable intelligence. Looking ahead, we propose interesting possibilities for further research. To achieve even higher accuracy and resilience, studying ensemble methods which combine various learning algorithms to take advantage of their combined strengths holds promise. Furthermore, exploring deep learning architectures especially those skilled at handling intricate, high-dimensional network traffic data may allow IDS capabilities to be further enhanced. This research ultimately promotes the expansion of cybersecurity defences by developing a deeper knowledge of IDS approaches and their underlying algorithms. This gives organisations the capacity to adjust and preserve their resilience in the face of a constantly evolving threat landscape in intricate IT environments. Check out the following link to view the source code of our study https://github.com/Chandini0209/IDS-using-ML-algorithms