Abstract:
Card fraud costs cardholders and issuers a lot. Modern methods detect fraud from transaction records using machine learning. Manually generating features requires domain knowledge and may be behind fraud’s modus operandi, so the online detection system must automatically focus on the most relevant fraudulent behavior patterns. For credit card fraud detection, we propose a spatial-temporal attention-based graph network (STAGN). A graph neural network first learns temporal and location-based transaction graph features. We then use spatial-temporal attention to feed learned tensor representations into a 3D convolution network. 3D convolution and detection networks learn attentional weights end-to-end. We then run extensive experiments on the real-world card transaction dataset. STAGN outperforms state-of-the-art baselines in AUC and precision-recall curves. We also conduct empirical studies with domain experts on the proposed fraud detection and knowledge discovery method, which shows its superiority in detecting suspicious transactions, mining spatial and temporal fraud hotspots, and uncovering fraud patterns. The method also works in other user behavior-based tasks. Finally, to address big data challenges, we integrate our proposed STAGN into the fraud detection system as the predictive model and present the implementation details of each module.
Note: Please discuss with our team before submitting this abstract to the college. This Abstract or Synopsis varies based on student project requirements.
Did you like this final year project?
To download this project Code with thesis report and project training... Click Here
