ISCX 2012 Dataset: Anomaly Detection via Bayesian Optimization & ML

This report presents an anomaly detection framework leveraging Bayesian Optimization (BO) to optimize parameters for machine learning algorithms, specifically Support Vector Machine with Gaussian kernel (SVM-RBF), Random Forest (RF), and k-Nearest Neighbor (k-NN). The study addresses the increasing prevalence of network attacks and the need for robust intrusion detection systems. BO is employed to identify optimal model parameters for classifying network intrusions, and the performance of the proposed methods is evaluated using the ISCX 2012 dataset. The experimental results demonstrate promising outcomes in terms of accuracy, precision, recall, and false alarm rate, highlighting the effectiveness of the BO-optimized machine learning approach for network anomaly detection. Desklib provides access to similar solved assignments and resources for students.