Fraud phone recognition method and system based on auto-encoder

Abstract

The invention discloses a fraud phone recognition method and system based on an auto-encoder. The method comprises the following steps: collecting detailed call data and user portrait data of a fraud phone user and a non-fraud phone user in a certain period of time; extracting dynamic communication features and static portrait features for each user; carrying out feature fusion on the communication features and the portrait features, selecting effective features, and carrying out normalization operation on the effective features; establishing an auto-encoder model under the integrated framework and training the model; setting an error threshold value according to the error distribution of the characteristics of the non-fraud phone samples participating in model training; performing abnormality diagnosis on the test set, and identifying a fraud call; according to the method, the fraud phone recognition problem is creatively regarded as the anomaly detection problem, and the problem of sample category imbalance can be effectively solved by adopting a method based on an auto-encoder; according to the self-encoder model under the integrated framework provided by the invention, relatively high accuracy, excellent generalization performance and relatively high training speed can be obtained.

Description

technical field [0001] The invention belongs to the technical field of telecommunication fraud identification, and in particular relates to a method and system for identifying fraudulent calls based on an autoencoder. Background technique [0002] In recent years, with the rapid development of the Internet and the widespread popularization of mobile communication devices, telephone not only brings great convenience to people's daily work and life, but also greatly increases the probability of telecommunication fraud. In phone scams, criminals typically use a single ID to apply for multiple phone numbers, and then use those phone numbers to make numerous calls, spread false information, lure others into transferring money, and so on. Telephone fraud not only causes the defrauded persons to suffer economic losses and great physical and mental harm, but also seriously endangers the normal social and financial order, frantically devours the social foundation of integrity, create...

AI Technical Summary

Problems solved by technology

[0004] Both of the above two technical routes are extremely unbalanced in categories, resulting in low recall and precision of model recognition

Images