Fraud phone dynamic identification interval construction method

Abstract

The invention discloses a fraud phone dynamic identification interval construction method. According to the method, a mode of combining hyper-parameter optimization and the gradient elevator is innovatively proposed to construct the fraud phone recognition model, the hyper-parameter optimization algorithm is utilized to optimize the parameters of the gradient elevator, and the model recognition effect is improved. According to the method, a random forest algorithm is used for carrying out data feature selection, and dimensions with feature importance greater than 0.8 are selected to construct fraud phone feature vectors. The user ticket data is typical unbalanced data, the invention proposes to sample the data by using an undersampling and oversampling combined mixed sampling method, the imbalance of data distribution is relieved, and the method is a feasible method through experimental verification. The invention provides a parameterization method based on a probability prediction model, which takes the probability output by a classifier as the confidence coefficient of a sample, and constructs a fraud phone dynamic recognition interval according to the confidence coefficient of the sample output by the model.

Description

technical field [0001] The invention relates to the field of Internet communication and the field of artificial intelligence, and is a method for constructing a dynamic identification interval of fraudulent calls, which can be applied to the field of telecommunications anti-fraud. Background technique [0002] Fraud phone calls have seriously disrupted the normal communication order, hindered citizens' freedom of communication, and interfered with the normal work and life of the people. It has become a serious problem in today's society. How to effectively identify and intercept fraudulent calls plays an important role in the telecommunications anti-fraud mechanism, which has attracted extensive attention from academia, industry and government funding agencies. [0003] In related technologies, it is a common method to identify fraudulent calls by crowdsourcing labeling, but the cost of crowdsourcing labeling is high and the efficiency is low. With the rapid development of ...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a method for constructing a dynamic interval for fraudulent call identification, which aims to solve the problem of low accuracy of fraudulent call identification in the anti-fraud scene in the telecommunications field, that is, telecom operators can use this model to complete fraudulent call identification and take Corresponding control measures to reduce user losses and improve user experience

Images