Detection of online phishing email using dynamic evolving neural network based on reinforcement learning

摘要: Abstract Despite state-of-the-art solutions to detect phishing attacks, there is still a lack of accuracy for the detection systems in online mode which leading loopholes web-based transactions. In this research, novel framework proposed combines neural network with reinforcement learning attacks first time. The model has ability adapt itself produce new email system that reflects changes newly explored behaviours, accomplished by adopting idea enhance dynamically over solves problem limited dataset automatically adding more emails offline mode. A algorithm explore any behaviours dataset. Through rigorous testing using well-known data sets, we demonstrate technique can handle zero-day high performance levels achieving accuracy, TPR, and TNR at 98.63%, 99.07%, 98.19% respectively. addition, it shows low FPR FNR, 1.81% 0.93% Comparison other similar techniques on same outperforms existing methods.