An Abnormal Phone Identification Model with Meta-learning Two-layer Framework Based on PCA Dimension Reduction
Yahan Yuan, Ke Ji, Runyuan Sun, Kun Ma, Zhenxiang Chen, Lin Wang · 2019
In the telecommunications industry, it is a critical and challenging problem that identify fraudulent calls in time. In the traditional abnormal phone identification method, there are generally cases where the initiative is weak and the recognition accuracy is low. In order to solve the problem of data sample imbalance and dirty data in the sample set, we use ensemble algorithms to improve the recognition accuracy of abnormal phones. Specially, we design a meta-learning two-layer framework (MTF) algorithm by integrating heterogeneous learners based on PCA dimension reduction. The experiment demonstrates that the MTF model has a great improvement in the abnormal phone identification compared with traditional classification method.