The development task's stated goal was to investigate potential uses of machine learning methods for anomaly and malicious pattern detection in network data. The performance of these models on several publicly accessible benchmark datasets was also to be evaluated. For this study, we used four datasets that stand in for various types of networks and attacks: NSL-KDD, UNSW-NB15, CICIDS2017, and ToN_IoT. The data was preprocessed using a thorough pipeline that included reduced dimensions, trained several ML and DL models, and then compared and evaluated their performance in binary and multiclass classification. We tested many different models, including k-Nearest Neighbors, Decision Tree, Multi-layer Perceptron, Random Forest, Support Vector Machine, Logistic Regression, and Long Short-Term Memory networks.