Ad Click Fraud Detection Using Machine Learning Algorithm

Abstract

Online advertising platforms have a serious problem with ad click fraud as it creates fraudulent clicks that skew campaign results and raise losses. This paper introduces an attention-based Convolutional Neural Network (CNN) framework for real-time ad click fraud detection in order to overcome the shortcomings of traditional machine learning and deep learning models in identifying intricate contextual patterns. In order to improve prediction reliability, the suggested improvement incorporates an attention mechanism into the CNN architecture to highlight the most instructive characteristics and sequential relationships in clickstream data. Furthermore, real-time fraud prediction via an interactive interface is made possible via a web-based deployment that makes use of the Flask framework. The results of experimental assessment show that the attention-enhanced CNN outperforms baseline models in terms of accuracy and resilience, underscoring its usefulness for scalable and realistic ad click fraud detection in dynamic online advertising contexts.