Unveiling Machine Learning Paradigms for Robust Malware Detection in Personal Data Security

In the world of computers and software, data is an asset; a personal, valuable, and important component in every person’s life. Malware used to extract hidden and personal data is slowly spreading around the globe. Conventional signature-based malware detection techniques are finding it difficult to keep up with the constant growth of malware. A viable substitute is machine learning (ML), which can recognize intricate patterns and adjust to new threats. This study compares 11 different Machine Learning models, to find the best model that can detect malware. The algorithms are trained and tested on a dataset called TUNADROMD with and without using Principle Component Analysis (PCA). The Random Forest is found to be the best-performing model with and without PCA with accuracies of 0.99 and 0.98 respectively. The worst-performing model is found to be Logistic Regression both with and without PCA with accuracies of 0.98 and 0.81 respectively.