Malicious URL detection Using majority vote method with machine learning and deep learning models
A.C Rakotoasimbahoaka, Iadaloharivola Randria, Nicolas Raft Razafindrakoto · 2020
Website can be accessed by URL (Uniform Resource Locator) and hackers use it with unsolicited content, which leads to cyber-attacks known as malicious URL and causes economic loss around the world. Machine learning, deep learning and a lot of other mechanisms detect these malicious URLs. However, the combination of these models faces over-fitting problem if they do not follow the same law (binomial and multinomial laws combined). This paper suggests majority vote system of different machine learning and deep learning models. With 420 464 URLs, result is interesting for combination of Random Forest (RF), Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN) models, giving an accuracy rate of 93% and solving over-fitting problem.