Anomaly Detection in Cybersecurity using Machine Learning Classifiers

This research examines anomaly detection in cybersecurity by utilizing various machine learning classifiers, including Naive Bayes, Artificial Neural Network (ANN), Support Vector Machine (SVM) and Random Forest, applied to the UGRansom dataset. The UGRansom dataset is a comprehensive resource for analyzing ransomware and zeroday attacks, containing essential data such as timestamps, attack type labels, protocol details, and extensive network flow records. It also categorizes ransomware families, provides insights into malware behavior, and estimates financial losses in both USD and bitcoins, making it an invaluable tool for cybersecurity research. By applying these classifiers, the study aims to effectively detect malicious activities in network traffic. The effectiveness of the classifiers will be investigated through the deployment of various metrics, including precision, accuracy, recall and F1-score thereby revealing their respective strengths and drawbacks concerning anomaly detection. The results are expected to offer profound insights into the functionalities of these algorithms for the instantaneous detection and mitigation of threats in the field of cybersecurity. Ultimately, the findings will support efforts to enhance network security and develop stronger defenses against evolving cyber threats, aiding organizations in protecting their digital assets from ransomware and other similar attacks.