Investigating the Efficacy of Hyper Parameter Tuned Machine Learning in Malware Detection

Abstract

This paper investigates the efficacy of hyperparameter tuning in enhancing machine learning models for malware detection. Given the escalating threats posed by sophisticated malware, traditional detection methods often fall short, necessitating more advanced and adaptive technologies. This study utilizes several popular machine learning algorithms—including decision trees, support vector machines (SVMs), and neural networks—optimized through rigorous hyperparameter tuning to maximize their detection capabilities.

The methodology centers on a comparative analysis of models pre and post hyperparameter adjustments, employing techniques such as grid search and random search to identify optimal configurations. The dataset comprises a diverse array of malware samples, ensuring comprehensive training and testing scenarios. Each model's performance is evaluated based on accuracy, precision, recall, and F1-score—metrics that collectively gauge the models' abilities to correctly identify malware without misclassifying benign applications.