eXpose Neural Network for Malicious URL Detection

This research article presents eXpose, a character-level convolutional neural network designed to detect malicious URLs, file paths, and registry keys. By automating feature extraction and classification, eXpose outperforms traditional manual feature extraction methods, achieving a 5%-10% higher detection rate at low false positive rates. The study highlights the potential of deep learning in enhancing cybersecurity detection capabilities.