A Self-Attention Network for Hierarchical Data Structures with an Application to Claims Management
This addresses fraud detection for insurance companies, but it is incremental as it applies known neural network methods to a specific domain problem.
The paper tackled fraud detection in health insurance claims by proposing self-attention and piecewise feed-forward neural networks to handle hierarchical data, showing that the self-attention method outperformed existing models on a dataset of two million claims.
Insurance companies must manage millions of claims per year. While most of these claims are non-fraudulent, fraud detection is core for insurance companies. The ultimate goal is a predictive model to single out the fraudulent claims and pay out the non-fraudulent ones immediately. Modern machine learning methods are well suited for this kind of problem. Health care claims often have a data structure that is hierarchical and of variable length. We propose one model based on piecewise feed forward neural networks (deep learning) and another model based on self-attention neural networks for the task of claim management. We show that the proposed methods outperform bag-of-words based models, hand designed features, and models based on convolutional neural networks, on a data set of two million health care claims. The proposed self-attention method performs the best.