Effective Phishing Website Detection Based on Improved BP Neural Network and Dual Feature Evaluation

Nowadays, phishing poses a big threat to people's daily network environment. By phishing, attackers obtain the network users private information by inducing them to open illegal websites. Due to the active learning ability and preferable classifying ability for many datasets, BP neural network is an important heuristic machine learning method in phishing websites detection and prevention. However, improper selection of initial parameters, such as the initial weight and threshold, will induce the BP neural network into local minimum and slow learning convergence. Aiming at these problems, this paper proposes DF.GWO-BPNN, an effective phishing website detection model based on the improved BP neural network and dual feature evaluation mechanism. Under this model, the grey wolf algorithm is firstly used to optimize the BP neural network to reasonably select initial parameters. Then, the dual feature mechanism is used to evaluate the results of the improved BP neural network. By the dual feature evaluation mechanism, the accuracy of phishing website recognition is improved. Meanwhile, the black and white list is used to improve the efficiency of the proposed model. The DF.GWO-BPNN model is compared with some existing phishing website detection models. The experimental results have demonstrated that our model is accurate and strong adaptability.