Email communication has now become a necessary conversation medium in our daily life. Particularly for the finance sector, communication by email represents a primary role in their businesses. So, it is necessary to classify emails based on their performance. Email phishing is one of the most serious Internet phenomena that make various difficulties to business class essentially to the finance sector. Furthermore, phishing emails are increasing at a dangerous rate in recent days. Hence, more environmentally friendly technology for phishing detection is required to manage the risk of phishing emails. This paper proposes an intelligent system for identifying phishing emails using enhanced recurrent convolutional neural networks (ERCNN). The system performs as new ability to have a web browser as an extension that notifies the user mechanically during detection of phishing emails. The whole system is based on a deep learning approach, in particular supervisory mastery. We chose the Convolutional Neural Network (CNN) due to its excellent overall performance in this category. Our conscience is looking for a higher overall performance classifier by analyzing phishing emails' potential and selecting aenhancedmixture of them to train the classifier. The results shows that proposed work get98.8% accuracy and a total of 26 features.
Key words: phishing email detection, classification, recurrent convolutional neural networks, convolutional neural networks (CNN).
|
