Malware Detection in Files and URL’s using Machine Learning

Abstract

The use of Support Vector Machine (SVM) to machine learning-based malware detection is the main goal of this study. Since malware threats are always changing, more advanced detection techniques are required. With the help of SVM, a potent classification algorithm, our project is able to precisely identify malware based on a range of attributes. By developing a strong and effective solution that efficiently identifies and reduces malware threats, the aim is to improve cybersecurity and support continuous efforts to protect computer systems and networks. Malware has been posing a serious threat to embedded systems in recent times, and traditional software solutions like antivirus and patching haven't been able to keep up with the sophisticated and constantly changing bad programmes. In this work, we present guardol, a hardware-enhanced architecture designed to identify online malware. Guardol is a hybrid technique that combines FPGA and CPU. Our method seeks to capture malware's malevolent behaviour, or high-level semantics. In order to do this, we first suggest the frequency-centric model for building features out of benign samples and known malware's system call patterns. We next create a machine learning strategy in FPGA to train a classifier with these features, utilising a multilayer perceptron. The trained classifier is applied at runtime to categorise the unknown data as benign or malicious, with In this day and age of advanced technology, the internet has been embraced by most people. And with it has come an increased risk of malevolent cyberattacks by cybercriminals. The attacks are completed.