An insurance claim settlement intelligent anti-fraud method and system with automatic feature intersection

Abstract

The invention relates to an insurance claim settlement intelligent anti-fraud method with automatic feature intersection. The method comprises a single-feature construction step, for a step of constructing feature libraries according to different data types and feature intersection, for a step of performing multi-layer feature cross calculation on the features in the feature library; a step of constructing the weight and the offset of each feature and constructing a network model , a step of constructing a cross network and a deep neural network which are parallel to each other on a stackinglayer; a step of taking the cross network and the deep neural network as a combination layer, then inputting the whole combination layer into a log function layer, and the training step of the model by adopting a general neural network BP back propagation algorithm. The invention also provides an insurance claim settlement intelligent anti-fraud system with automatic feature intersection. According to the method, the problem of cross feature construction in an existing depth model can be solved, automatic construction and optimization of cross features are achieved, and therefore manpower input in the feature construction process is reduced.

Description

technical field [0001] The invention relates to the technical field of automatic anti-fraud discrimination algorithms, in particular to an intelligent anti-fraud method and system for insurance claims with automatic feature intersection. Background technique [0002] At present, there are a large number of fraudulent activities in the field of insurance claims, and how to detect fraud has always been a difficult problem for insurance practitioners and the field of insurance technology. With the development of machine learning technology, various deep neural networks have begun to be applied in the field of insurance anti-fraud. Existing deep learning anti-fraud models require a large amount of artificial feature engineering to ensure the effectiveness of the anti-fraud model. Generally speaking, the analysis based on a single feature cannot realize the anti-fraud model, and multiple features need to be cross-combined to construct new features. . However, such cross-feature...

AI Technical Summary

Problems solved by technology

[0004] Based on this, it is necessary to provide an intelligent anti-fraud method and system for insurance claims with automatic feature crossing, which can solve the problem of cross feature construction in existing deep models, realize automatic construction and optimization of cross features, and reduce the number of problems in the feature construction process. Human input

Images