Published Paper Details:

In recent years, the exponential growth of data has revolutionized the landscape of information security, necessitating advanced data analysis systems tailored for big data. Traditional techniques struggle to cope with the sheer volume, variety, and velocity of data generated across networks, making cyber attack detection increasingly challenging. Intrusion Detection Systems (IDS) leveraging big data technology offer a promising avenue for accurate and efficient analysis. This paper presents a novel approach utilizing Spark Chi SVM model for cyber attack detection. Leveraging ChiSqSelector for feature selection and employing a Support Vector Machine (SVM) classifier on the Apache Spark big data platform, the model demonstrates high performance, reduced training time, and efficiency for processing big data. Through experimentation with the KDD99[10] dataset, a comparison between Chi SVM and Chi logistic regression classifiers underscores the superiority of the SVM model. Additionally, the paper introduces a cutting-edge technology incorporating deep learning models for constructing a balanced representation of imbalanced raw datasets. By leveraging deep neural networks (DNN)[7] and decision tree (DT)[8] classifiers, this approach significantly enhances network attack detection. Validation using real-world critical infrastructure datasets showcases a remarkable improvement, with a 10% higher F1 score compared to conventional classifiers like Random Forests (RF)[6] and standard DNNs[7]. Achieving precision rates of 95.86% and 99.67% on datasets from gas pipeline and safe water treatment facilities respectively, this research paves the way for more accurate and effective cybersecurity measures.