Phishing Website Detection Using Novel Features And Machine Learning Approach
Phishing Website Detection Using Novel Features And Machine Learning Approach
复制标题
使用新颖功能和机器学习方法检测网络钓鱼网站
DOI：
--
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
et. al. S.T.Deepa
中科院分区：
文献类型：
--
作者：
et. al. S.T.Deepa
网络钓鱼，一种对个人产生不利影响的数字攻击，客户被协调到假冒网站，并被欺骗，以揭露他们的敏感和个人数据，包括记录密码，银行细节，ATM pin-card细节等，随后屏蔽敏感数据免受恶意软件或网络攻击；不规则林木钓鱼麻烦。从现有的网络钓鱼网站识别技术的限制来看，期望用户能够注意到并判断一个URL是网络钓鱼还是真实的，这是不合理的、浪费的和错误的。因此，为了解决这些困难，应该考虑采用机器人化的方法进行网络钓鱼站点识别。本研究旨在利用新的特征和机器学习算法来检测网络钓鱼网站。首先使用卷积自动编码器对输入的URL网站进行特征提取。然后将这些特征发送给深度神经网络分类器，以便更好地对钓鱼和合法URL进行分类。测试了该系统的准确率和检出率。结果表明，该系统对钓鱼网站的检测准确率最高，达到89%。
Phishing, a type of digital assault adversely affects individuals where the client is coordinated to counterfeit sites and tricked to uncover their delicate and individual data which incorporates passwords of records, bank subtleties, ATM pin-card subtleties and so forth Subsequently shielding touchy data from malwares or web real; irregular woods phishing is troublesome. Inferable from the restrictions of existing advances in identifying a phishing site, anticipating that the users should notice and can decide if a URL is phishing or genuine would be unreasonable, wasteful and mistaken. Consequently, in tending to these difficulties, a robotized approach should be considered for phishing site recognition. This research aims at detecting phishing website using novel features and machine learning algorithm. The input URL websites are first feature extracted using Convolutional auto encoder. Then those features are sent to deep neural network classifier for better classification of Phishing and legitimate URL’s. The system is tested for its accuracy and detection rate. It shows that the implemented system is best in detecting phishing websites with 89% accuracy.