IDN GAN AN INTELLIGENT DEEP NETWORK WITH GAN FOR ENHANCED CYBER THREAT DETECTION

Abstract

With cyber threats increasing at an alarming rate, security breaches have surged by 67% over the past five years, and cybercrime is expected to cost the world $10.5 trillion annually by 2025. Traditional manual threat detection methods rely on rule-based systems and human expertise, which are prone to errors, slow in response, and inefficient in handling large-scale data. To address these challenges, we propose a machine learning-based behavioral analysis approach for security threat detection, leveraging deep learning for improved accuracy. The proposed method begins with data preprocessing to clean and normalize the Security Threat Dataset, which contains four attack labels: Denial of Service (DoS), Probe, Remote-to-Local (R2L), and User-to-Root (U2R). To handle class imbalance, we employ the ADASYN (Adaptive Synthetic Sampling) technique, which generates synthetic minority class samples, ensuring a balanced dataset. The dataset is then split into training and testing sets to evaluate model performance. We compare traditional classifiers such as Naïve Bayes and Support Vector Machine (SVM) with our proposed Convolutional Neural Network with Generative Adversarial Network (CNN-GAN) based model, which captures complex patterns and hierarchical representations in network traffic data. Experimental results demonstrate that CNN-GAN outperforms existing methods in terms of detection accuracy, recall, and precision, making it a robust solution for real-time security threat analysis.