Phishing is a critical internet hazard and phishing losses progressively and it is caused by electronic means to deprive the users of sensitive information. Feature engineering is remaining essential for website-detection phishing solutions, although the quality of detection depends ultimately on previous knowledge of its features. Moreover, while the functionalities derived from different measurements are more precise, these characteristics take a lot of time to remove. This suggest a multidimensional approach to the detection of phishings focused on a quick detection mechanism through deep learning to overcome these limitations. The first step is to extract and use the character sequence features of the given URL for rapid classification through in-depth learning; this step does not include support from third parties or previous experience in phishing. It combine statistical URLs, web page code functions, website text features and easily categorise Profound learning in the second level on multidimensional functions. By the approach, the detection time of the threshold is shortened. The experimental results show that a rational adjustment of the threshold allows for the efficiency of the detection.
Analysis of Phishing Website Detection Using CNN and Bidirectional LSTM
05.11.2020
204102 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Phishing How Microsoft is fighting back against phishing
British Library Online Contents | 2006
Unidirectional and Bidirectional LSTM Models for Short-Term Traffic Prediction
DOAJ | 2021
|