Information push method, device and computer equipment based on confrontation network

Abstract

The invention discloses an information push method, device, computer equipment and storage medium based on an adversarial network, which involves artificial intelligence, including first repeating any of the kth grouping of the virtual object index matrix grouping and the k+1th grouping of the virtual object index matrix grouping. Obtain a virtual object index matrix and the grouping numbers corresponding to the obtained two virtual object index matrices to train the cycle-GAN model to be trained, and stop obtaining the virtual object index matrix from the classification result of the virtual object index matrix until it converges, Obtaining a cycle-GAN model; inputting the current virtual object index matrix into the cycle-GAN model for calculation to obtain a migration matrix corresponding to the current virtual object index matrix. Through the cycle-GAN model, the migration matrix similar to the current virtual object index matrix can be automatically found in a large number of virtual object index matrices as a parameter reference, without manual search, comparison and analysis of data, which improves the acquisition of recommended data efficiency.

Description

technical field [0001] The present invention relates to the technical field of intelligent decision-making of artificial intelligence, and in particular to an information push method, device, computer equipment and storage medium based on an adversarial network. Background technique [0002] At present, in the field of information distribution, there are more and more information recommendation algorithms. For example, a common method of information recommendation is to first obtain the user tags of the target user, and then push the target content to the target user after screening the target content corresponding to the user tag in the content library. The information recommendation algorithm in the prior art is suitable for scenarios where user tags are easy to obtain. Once the user’s user tags are difficult to obtain (for example, only the user’s high-dimensional feature matrix can be extracted), the existing information recommendation algorithm cannot be used. recommen...

AI Technical Summary

Problems solved by technology

[0004] The embodiment of the present invention provides an information push method, device, computer equipment, and storage medium based on an adversarial network, aiming to solve the problem that the user's user tag is difficult to obtain in the prior art, and it is necessary to obtain an approximate characteristic matrix with the user or enterprise When using a target feature matrix for information recommendation, a supervised model is generally used for screening or manual judgment, resulting in long acquisition time, low efficiency, and low accuracy of the target feature matrix.

Images