Malicious URLs and QR Code Classification Using Machine Learning and Deep Learning Techniques

Attackers commonly employ malicious URLs and harmful QR codes to spread malware and phishing scams; therefore, they need to be classified. Malicious URLs that lead visitors to phishing or malware-infected websites that steal personal data can be included in emails, social media posts, or website content. Rogue QR codes can be used to propagate malware, steal data, or direct users there, much like malicious websites can. In this paper, the classification of malicious content is provided using two approaches:(1) Malware URL categorization based on ML and malware QR codes (2) classification of malicious QR codes based on deep learning. The first technique classifies URLs as dangerous or benign using machine learning models that are trained on features derived from the URLs. Several ML techniques, such as Random Forest, Naive Bayes, and Support Vector Machine, are used to evaluate the model's performance. The second option focuses on classifying harmful QR codes using deep learning techniques. The classification of QR codes as malicious or benign is done using CNN and well-known transfer learning models like RESNET. In general, the offered approaches provide effective techniques to classify unsafe material, which can be utilized to enhance security.