Detecting File Types Using Machine Learning Algorithms

In this study, experiments have been carried out to analyze the file content and classify the type of file using machine learning algorithms. First a new benchmark dataset was created including popular file types used as malicious software. Then the proposed system was trained using different machine learning algorithms. The accuracy and runtime performance comparison of models were evaluated. As a result, the highest accuracy rate was obtained as 97.83% by XGBoost algorithm.

I. Introduction

Digital information is stored as files and each computer file is associated with a file type on storage medium. The detection of the file types correctly can be guide for identifying harmful and risky files in advance, and it leads to take necessary precautions without damaging the systems. Accurate detection of a file type has great importance in operating systems, firewalls, intrusion detection systems, antiviruses, forensics and many other areas [1].