Auto insurance fraud identification based on a CNN-LSTM fusion deep learning model
Huosong Xia, Yanjun Zhou, Zuopeng Zhang
Abstract
The traditional auto insurance fraud identification method relies heavily on feature engineering and domain knowledge, making it difficult to accurately and efficiently identify fraud when the amount of claim data is large and the data dimension is high. Deep learning models have strong generalisation abilities and can automatically complete feature extraction. This paper proposes a deep learning model for auto insurance fraud identification by combining convolutional neural network (CNN), long- and short-term memory (LSTM), and deep neural network (DNN). Our proposed method can extract more abstract features and help avoid the complex feature extraction process that is highly dependent on domain experts in traditional machine learning algorithms. Experiments demonstrate that our method can effectively improve the accuracy of auto risk fraud identification.