Classification model generation method based on missing data

Abstract

The invention discloses a classification model generation method based on missing data. The method comprises the steps of obtaining a missing data feature matrix; representing the missing data feature matrix by using a probability density function of a Gaussian mixture model to obtain a first feature matrix; obtaining a one-hot vector matrix according to a label corresponding to the data missing type; inputting the first feature matrix and the one-hot vector matrix into a hypergraph convolutional network model to enable the hypergraph convolutional network model to generate a prediction model based on missing data after training; specifically, a hypergraph convolutional network model calculates an expected response of an RELU neuron according to a first feature matrix to obtain a first hidden feature, and then obtains a first prediction label according to the first hidden feature; the hypergraph convolutional network model performs label propagation according to the one-hot vector matrix to obtain a second prediction label; and joint learning is carried out on the first prediction label and the second prediction label to generate a missing data-based classification model. By adopting the method, the model precision can be improved.

Description

technical field [0001] The invention relates to the technical field of neural networks, in particular to a method for generating a classification model based on missing data. Background technique [0002] In the field of deep learning, it is difficult to obtain complete data for certain application scenarios, such as personal privacy information in social networks, lost data of industrial sensors, etc. Therefore, when it is difficult to obtain complete data, how to complete the missing data is a very important issue for the training process of the model. There are currently three types of missing data completion methods: [0003] The first category is to fill in missing data using traditional methods. Imputation methods are widely used for data completion, such as mean imputation, matrix completion for matrix factorization and singular value decomposition (SVD), and multiple imputation. The second category is to use machine learning to estimate missing values, such as k-N...

AI Technical Summary

Problems solved by technology

[0004] However, although the above existing traditional methods and machine learning methods complement the missing features to a certain extent, they cannot achieve end-to-end training for the model; while the Gaussian mixture model represents the missing features. The graph convolutional network method ignores Therefore, in the process of training the model, the accuracy of the model generated by using the above three types of missing data completion methods is not high.

Images