Collaborative filtering recommendation algorithm based on variational inference
PurposeThe purpose of this paper is to alleviate the problem of poor robustness and over-fitting caused by large-scale data in collaborative filtering recommendation algorithms.Design/methodology/approachInterpreting user behavior from the probabilistic perspective of hidden variables is helpful to...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Tsinghua University Press
2020-03-01
|
| Series: | International Journal of Crowd Science |
| Subjects: | |
| Online Access: | https://www.sciopen.com/article/10.1108/IJCS-10-2019-0030 |
| _version_ | 1797986228804517888 |
|---|---|
| author | Kai Zheng Xianjun Yang Yilei Wang Yingjie Wu Xianghan Zheng |
| author_facet | Kai Zheng Xianjun Yang Yilei Wang Yingjie Wu Xianghan Zheng |
| author_sort | Kai Zheng |
| collection | DOAJ |
| description | PurposeThe purpose of this paper is to alleviate the problem of poor robustness and over-fitting caused by large-scale data in collaborative filtering recommendation algorithms.Design/methodology/approachInterpreting user behavior from the probabilistic perspective of hidden variables is helpful to improve robustness and over-fitting problems. Constructing a recommendation network by variational inference can effectively solve the complex distribution calculation in the probabilistic recommendation model. Based on the aforementioned analysis, this paper uses variational auto-encoder to construct a generating network, which can restore user-rating data to solve the problem of poor robustness and over-fitting caused by large-scale data. Meanwhile, for the existing KL-vanishing problem in the variational inference deep learning model, this paper optimizes the model by the KL annealing and Free Bits methods.FindingsThe effect of the basic model is considerably improved after using the KL annealing or Free Bits method to solve KL vanishing. The proposed models evidently perform worse than competitors on small data sets, such as MovieLens 1 M. By contrast, they have better effects on large data sets such as MovieLens 10 M and MovieLens 20 M.Originality/valueThis paper presents the usage of the variational inference model for collaborative filtering recommendation and introduces the KL annealing and Free Bits methods to improve the basic model effect. Because the variational inference training denotes the probability distribution of the hidden vector, the problem of poor robustness and overfitting is alleviated. When the amount of data is relatively large in the actual application scenario, the probability distribution of the fitted actual data can better represent the user and the item. Therefore, using variational inference for collaborative filtering recommendation is of practical value. |
| first_indexed | 2024-04-11T07:29:37Z |
| format | Article |
| id | doaj.art-4ce2d6041a4c4f13b829e17a090f31d5 |
| institution | Directory Open Access Journal |
| issn | 2398-7294 |
| language | English |
| last_indexed | 2024-04-11T07:29:37Z |
| publishDate | 2020-03-01 |
| publisher | Tsinghua University Press |
| record_format | Article |
| series | International Journal of Crowd Science |
| spelling | doaj.art-4ce2d6041a4c4f13b829e17a090f31d52022-12-22T04:36:56ZengTsinghua University PressInternational Journal of Crowd Science2398-72942020-03-0141314410.1108/IJCS-10-2019-0030Collaborative filtering recommendation algorithm based on variational inferenceKai Zheng0Xianjun Yang1Yilei Wang2Yingjie Wu3Xianghan Zheng4College of Mathematics Computer Science/College of Software, Fuzhou University, Fuzhou, ChinaCollege of Mathematics Computer Science/College of Software, Fuzhou University, Fuzhou, ChinaCollege of Mathematics Computer Science/College of Software, Fuzhou University, Fuzhou, ChinaCollege of Mathematics Computer Science/College of Software, Fuzhou University, Fuzhou, ChinaCollege of Mathematics Computer Science/College of Software, Fuzhou University, Fuzhou, ChinaPurposeThe purpose of this paper is to alleviate the problem of poor robustness and over-fitting caused by large-scale data in collaborative filtering recommendation algorithms.Design/methodology/approachInterpreting user behavior from the probabilistic perspective of hidden variables is helpful to improve robustness and over-fitting problems. Constructing a recommendation network by variational inference can effectively solve the complex distribution calculation in the probabilistic recommendation model. Based on the aforementioned analysis, this paper uses variational auto-encoder to construct a generating network, which can restore user-rating data to solve the problem of poor robustness and over-fitting caused by large-scale data. Meanwhile, for the existing KL-vanishing problem in the variational inference deep learning model, this paper optimizes the model by the KL annealing and Free Bits methods.FindingsThe effect of the basic model is considerably improved after using the KL annealing or Free Bits method to solve KL vanishing. The proposed models evidently perform worse than competitors on small data sets, such as MovieLens 1 M. By contrast, they have better effects on large data sets such as MovieLens 10 M and MovieLens 20 M.Originality/valueThis paper presents the usage of the variational inference model for collaborative filtering recommendation and introduces the KL annealing and Free Bits methods to improve the basic model effect. Because the variational inference training denotes the probability distribution of the hidden vector, the problem of poor robustness and overfitting is alleviated. When the amount of data is relatively large in the actual application scenario, the probability distribution of the fitted actual data can better represent the user and the item. Therefore, using variational inference for collaborative filtering recommendation is of practical value.https://www.sciopen.com/article/10.1108/IJCS-10-2019-0030collaborative filteringprobabilistic perspectivevariational inferencevariational auto-encoderkl-vanishing problem |
| spellingShingle | Kai Zheng Xianjun Yang Yilei Wang Yingjie Wu Xianghan Zheng Collaborative filtering recommendation algorithm based on variational inference International Journal of Crowd Science collaborative filtering probabilistic perspective variational inference variational auto-encoder kl-vanishing problem |
| title | Collaborative filtering recommendation algorithm based on variational inference |
| title_full | Collaborative filtering recommendation algorithm based on variational inference |
| title_fullStr | Collaborative filtering recommendation algorithm based on variational inference |
| title_full_unstemmed | Collaborative filtering recommendation algorithm based on variational inference |
| title_short | Collaborative filtering recommendation algorithm based on variational inference |
| title_sort | collaborative filtering recommendation algorithm based on variational inference |
| topic | collaborative filtering probabilistic perspective variational inference variational auto-encoder kl-vanishing problem |
| url | https://www.sciopen.com/article/10.1108/IJCS-10-2019-0030 |
| work_keys_str_mv | AT kaizheng collaborativefilteringrecommendationalgorithmbasedonvariationalinference AT xianjunyang collaborativefilteringrecommendationalgorithmbasedonvariationalinference AT yileiwang collaborativefilteringrecommendationalgorithmbasedonvariationalinference AT yingjiewu collaborativefilteringrecommendationalgorithmbasedonvariationalinference AT xianghanzheng collaborativefilteringrecommendationalgorithmbasedonvariationalinference |