In last decades, the web and online services have revolutionizedthe modern world. However, by increasing our dependence ononline services, as a result, online security threats are also increas-ing rapidly. One of the most common online security threats isa so-called Phishing attack, the purpose of which is to mimic alegitimate website such as online banking, e-commerce or socialnetworking website in order to obtain sensitive data such as user-names, passwords, financial and health-related information frompotential victims. The problem of detecting phishing websites hasbeen addressed many times using various methodologies fromconventional classifiers to more complex hybrid methods. Recentadvancements in deep learning approaches suggested that the clas-sification of phishing websites using deep learning neural networksshould outperform the traditional machine learning algorithms.However, the results of utilizing deep neural networks heavily de-pend on the setting of different learning parameters. In this paper,we propose a swarm intelligence based approach to parameter set-ting of deep learning neural network. By applying the proposedapproach to the classification of phishing websites, we were ableto improve their detection when compared to existing algorithms.