NEURAL GUARDS: SAFEGUARDING CYBERSPACE WITH FEED-FORWARD NEURAL NETWORKS IN MALICIOUS WEBSITE DETECTION

Abstract

The internet has become a vulnerable platform due to an increase in malicious websites. As millions of users access online services daily, hackers continuously launch attacks, resulting in financial, personality, and malware thefts. To address this serious threat, machine learning algorithms such as support vector machines (SVM) have been used to identify and flag malicious websites. This paper proposes a robust model for detecting malicious websites using a feed-forward neural network (FFNN) algorithm. The model was trained using a dataset of 48,006 legitimate websites and an equal number of malicious websites to achieve an accuracy level of 97%.The process of deriving features from a URL plays an integral role in the model's ability to identify malicious websites. Lexical features, such as the number of dots and the length of the url, were used to prepare the dataset for training. The FFNN algorithm was then applied to the dataset, which resulted in the creation of a deep-learning model that was deployed using the Flask framework to enable users to enter website URLs for detection.

The proposed model offers a robust tool for detecting malicious websites with a high degree of accuracy. The model provides a promising solution for addressing the risks associated with malicious websites and can serve as a foundation for further study and implementation using other machine learning algorithms.