Audio recommendation model training method and device, medium and equipment

Abstract

The invention provides an audio recommendation model training method and device, a computer readable storage medium and electronic equipment, and relates to the technical field of artificial intelligence. The method comprises the following steps: acquiring M groups of sample data, wherein each group of sample data comprises target music and N pieces of candidate music after a user plays the target music; obtaining the actual probability that the user plays the ith candidate music in the jth group after playing the target music in the jth group, and obtaining the actual probability distribution of the music in the jth group; based on a to-be-optimized model parameter of an audio recommendation model, determining a simulation probability that the user plays the ith candidate music in the jth group after playing the target music in the jth group, and obtaining simulation probability distribution about the music in the jth group; and determining a loss function according to the actual probability distribution and the simulated probability distribution, and optimizing model parameters based on the loss function to obtain a trained audio recommendation model. According to the scheme, the exposure rate of the music can be improved, and the prediction accuracy of the recommended music can be improved.

Description

technical field [0001] The present disclosure relates to the technical field of artificial intelligence, and in particular, to a method and apparatus for training an audio recommendation model, as well as a computer-readable storage medium and an electronic device for implementing the above method. Background technique [0002] In the scenario of playing music, the music application generally provides the user with a playlist (listen to your heart, music streaming, playlist recommendation, etc.) to facilitate the user to listen. With the development of artificial intelligence technology, user preferences are more and more taken into account in the process of determining recommended playlists. [0003] In the related art, a music sequence of a user listening to a song is obtained, and a music representation is obtained indirectly by means of ideas such as Word2vec, and then a music recommendation model is trained to recommend music to the user. However, although the songs re...

AI Technical Summary

Problems solved by technology

However, although the songs recommended by this music recommendation scheme can describe the user's listening behavior, since user behavior is generally shown in the head of popular music, the songs recommended by this scheme generally cannot contain information that lacks user behavior. music

[0004] It can be seen that the audio recommendation scheme provided by related technologies has the problem of low music exposure

Images