A Novel IDS Framework Combining PCA and Random Forest
Authors:-Roshan Kumar S, Mageshwaran S D, Krithik S R, Dr. U. Surendar
Abstract-The aim of this project is to develop an application capable of identifying the type of attack on a system and detecting intruders using an intrusion detection system. [1] [6] Various machine learning techniques have been previously applied to IDS to enhance intrusion detection accuracy. This research introduces a novel approach by integrating principal component analysis and the random forest classification algorithm to build a more efficient IDS. PCA helps refine the dataset by reducing its dimensionality, simplifying data processing, while the random forest algorithm ensures precise classification of network activities. Experimental results indicate that this method outperforms traditional techniques such as support vector machine, naïve bayes and decision tree, delivering high accuracy. IDS functions as a network security measure, identifying threats, malicious activities, and attack types. A key limitation in conventional IDS is that if the primary detector fails, intrusions remain undetected. To address this, the proposed system implements multiple detection mechanisms, ensuring that if one detector fails, other continue monitoring and identifying threats, thereby, strengthening system security.
