Optimizing Intrusion Detection Systems: Leveraging Machine Learning and Deep Learning with Advanced Parameter Tuning
Amit Kumar, Sachin Ahuja, Ganesh Gupta · 2025
The paper explores the effectiveness of many machine learning methods in IDS (Intrusion detection systems). This paper test and analyze the effectiveness of PCA, Decision Tree, Neural Network and Random Forest algorithms for the identification of malicious activity in system logs or network traffic. Our findings suggest that Decision Tree, Random Forest, and PCA can identify intrusion attacks quite well, achieving 98% accuracy. In turn, Neural Network’s performance also achieves 99.5% accuracy. Such a study presents promising potential of PCA, Random Forest, and Decision Tree algorithms for robust intrusion detection and serves as promising paths toward practical implementation of enhancing cyber-security protocols. Key words—Intrusion, machine learning, cyber-security, TCP, UDP, Neural Network