Neural network model training and object recommendation methods and apparatuses

Abstract

The invention relates to the technical field of computers, in particular to neural network model training and object recommendation methods and apparatuses. The neural network model training method comprises the steps of determining an eigenvector in a one-hot form corresponding to sample data, for each piece of sample data in multiple pieces of pre-collected sample data; according to the eigenvectors in the one-hot forms, corresponding to the sample data and sample tags of the sample data, training an LR model; according to the trained LR model, determining N weight values corresponding to Ndata values; for the sample data, selecting target weight values corresponding to the data values 1 in the eigenvectors in the one-hot forms, corresponding to the sample data from the N weight values,thereby obtaining M target weight values, wherein M is less than N; and inputting the M target weight values of the sample data, as new eigenvalues of the sample data, to a neural network model for training the neural network model. Therefore, the efficiency of training the neural network model can be improved.

Description

technical field [0001] The present application relates to the field of computer technology, in particular to a neural network model training and object recommendation method and device. Background technique [0002] In the traditional technology, after the sample data is collected, the neural network model is trained directly according to the sample data and the sample labels of the sample data. However, the sample data collected above usually includes information of multiple dimensions, which leads to low efficiency of neural network model training, and the trained neural network model is relatively complex. Due to the complexity of the neural network model, when it is applied online, it will take a long time to recommend the target object to the user, that is, its timeliness is relatively poor, which will cause the neural network model to be unable to be applied to the comparison of timeliness requirements. high scene. Contents of the invention [0003] This applicatio...

AI Technical Summary

Problems solved by technology

However, the sample data collected above usually includes information of multiple dimensions, which will lead to low efficiency of neural network model training, and the trained neural network model is relatively complex.

Due to the complexity of the neural network model, when it is applied online, it will take a long time to recommend the target object to the user, that is, its timeliness is relatively poor, which will cause the neural network model to be unable to be applied to the comparison of timeliness requirements. high scene

Images