Research on Building Bank Anti-fraud Model Based on Tri-training Semi-Supervised Learning and Fuzzy SVM Active Learning
Xiaoguo Wang, Luxi Liu · 2017
With the rapid development of Internet finance and its applications, online banking fraud is becoming increasingly frequent.How to accurately identify the fraud data among the huge amount of transactions, is the urgent needs of Third Party Payment Center, Channel Department and other departments of banks.As to this problem, this paper proposes a recognition method based on tri-training semi-supervised learning and fuzzy SVM active learningaiming at researching thathow to build aneffectiveanti-fraud model for banks.Experimental results show that the method has encouraging recognition accuracy, which provides an effective scheme for banks' anti-fraud model training, building and fraud recognition. Active learning and fuzzy SVMActive learning can filter out unlabeled data that has more information and is more useful for model's training to be labeled by professors, raising the number of the labeled data.Then the new 374