Evaluation of Machine Learning Techniques in the Classification of Malware Attacks

This research examines the performance of Random Forest and Decision Tree algorithms in classifying a dataset defined by various process metrics, such as hash, categorisation, and memory management parameters like total_vm, shared_vm, and maj_flt. To ensure the accuracy and reliability of the data, the dataset underwent comprehensive preprocessing. Both algorithms were trained and tested, achieving strong classification accuracy, indicating their suitability for the given dataset. The Random Forest algorithm, an ensemble learning method, showed strong performance by effectively managing the dataset's complexities through the use of multiple decision trees, which enhanced generalization. In contrast, the Decision Tree algorithm, valued for its interpretability, successfully pinpointed key features that influenced the classifications, offering insightful views into the data's underlying patterns. The models' performance was thoroughly evaluated using key metrics, including accuracy, precision, recall, and F1 score, ensuring a comprehensive assessment of their effectiveness. Both algorithms were effective in reducing false positives and false negatives, highlighting their reliability for practical use. While both models achieved high accuracy, the choice between them may depend on the need for either interpretability or superior performance. This study contributes to the understanding of machine learning applications in process classification, stressing the importance of selecting the right model based on dataset characteristics and the need for continuous evaluation in real-world applications.