Model training and abnormal account recognition method and device

Abstract

The invention discloses a model training and abnormal account recognition method and device. The method comprises the following steps: firstly, obtaining a pre-labeled first account and an unlabeled second account as training samples; secondly, determining each node in the heterogeneous network according to the account data of each user account; establishing a heterogeneous network according to the incidence relation among the nodes; determining the node attribute of each node in the heterogeneous network; inputting the heterogeneous network and the node attribute of each node into a to-be-trained account identification model; and determining a feature vector and a prediction probability of each user account, finally determining a loss function according to the prediction probability of each first account, the annotation information, the feature vector of each first account and the prediction probability of each second account, and adjusting model parameters by taking the minimum lossfunction as an optimization target so as to identify an abnormal account through the account identification model. Model training is carried out in a semi-supervised mode, so that the labeling time and labeling cost of the user account are reduced, and the recognition cost of the abnormal account is saved.

Description

technical field [0001] The present application relates to the field of Internet technology, in particular to a method and device for model training and identification of abnormal accounts. Background technique [0002] With the development of Internet technology, more and more users perform various businesses through network platforms, such as financial management and transactions through financial platforms, chatting and comments through social platforms, etc. Due to the characteristics of anonymity and simple information verification of the Internet platform, some users use their user accounts on the Internet platform to conduct some improper activities, such as: conducting illegal activities such as money laundering on financial platforms, and publishing improper remarks on social networks. In order to prevent improper behavior on the network platform, the network platform can control by identifying abnormal accounts on the platform, blocking abnormal accounts, restrictin...

AI Technical Summary

Problems solved by technology

[0004] The embodiment of this specification provides a method and device for model training and identification of abnormal accounts, which is used to partially solve the problem that in the prior art, the training process of the multi-layer perceptron requires a large number of pre-marked training samples, which takes a long time and costs a lot. , leading to the problem of high cost of abnormal account identification

Images