Design And Implementation Of Music Recommendation Algorithm Based On Generative Adversarial Networks

With the further development of the Internet and the rise of the mobile Internet,the number of users has increased significantly,resulting in more and larger capacity of user data.In order to solve the problem of how to use this data to provide more accurate services,the recommender system came into being.In recent years,although there have been many recommendation algorithms in the fields of film,news and books,there are not many recommender systems in the music field,mainly because the data in this field is often not public,and the scores we use are different.Basically no public data sets mention user ratings.This paper is based on the Echo Nest dataset(one of datasets in MSD)and the IRGAN information retrieval network model.By comparing the SVD and BiasSVD matrix factorization without using the anti-network and the SVD and BiasSVD matrix factorization models using the adversarial neural network,I hope to get from the adversarial neural network.Find ways to improve the performance of the current recommended model.And by comparing the SVD matrix factorization model using the adversarial neural network and the model using the MLP network,it is judged whether the adversarial neural network has sufficient ability to extract hidden features.Finally,the experimental results show that the adversarial network does have the recommended effect of improving the SVD and BiasSVD models,but the feature extraction effect is still inferior compared with MLP.