研讨会主题 Graph Machine Learning Practices on Financial Fraud Detection in Post-Pandemic Era 研讨会时间 Date: 13 September, 2022 (Tue.)Time: 12:00 - 13:00 pm HKTLanguage: English 研讨会简介 In recent years, graph machine learning methods, especially graph neural network, have achieved remarkable successes in fraud detection in digital financial platform due to their powerful feature extraction and information aggregation capabilities. However, special challenges such as class imbalance or adversarial attack from fraudsters has become important challenges for the further improvement of graph machine learning method-based solutions. How to address the class-imbalance issues in GNN field, how to purify the adversarial attacks in the graph structure and how to enhance the robustness of GNN to improve the generalization ability of out-of-distribution samples have become recent research hotspots. In this talk, I will introduce the research progresses of our research gr
………………………………