Bounded matrix factorization for recommender system

被引:51
作者
Kannan, Ramakrishnan [1 ]
Ishteva, Mariya [2 ]
Park, Haesun [1 ]
机构
[1] Georgia Inst Technol, Sch Computat Sci & Engn, Atlanta, GA 30338 USA
[2] Vrije Univ Brussel, Dept ELEC, Brussels, Belgium
基金
欧洲研究理事会; 美国国家科学基金会;
关键词
Low-rank approximation; Recommender systems; Bound constraints; Matrix factorization; Block coordinate descent method; Scalable algorithm; Block; CONSTRAINED LEAST-SQUARES; NONNEGATIVE MATRIX; ALGORITHMS;
D O I
10.1007/s10115-013-0710-2
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Matrix factorization has been widely utilized as a latent factor model for solving the recommender system problem using collaborative filtering. For a recommender system, all the ratings in the rating matrix are bounded within a pre-determined range. In this paper, we propose a new improved matrix factorization approach for such a rating matrix, called Bounded Matrix Factorization (BMF), which imposes a lower and an upper bound on every estimated missing element of the rating matrix. We present an efficient algorithm to solve BMF based on the block coordinate descent method. We show that our algorithm is scalable for large matrices with missing elements on multicore systems with low memory. We present substantial experimental results illustrating that the proposed method outperforms the state of the art algorithms for recommender system such as stochastic gradient descent, alternating least squares with regularization, SVD++ and Bias-SVD on real-world datasets such as Jester, Movielens, Book crossing, Online dating and Netflix.
引用
收藏
页码:491 / 511
页数:21
相关论文
共 29 条
  • [21] Projected gradient methods for nonnegative matrix factorization
    Lin, Chih-Jen
    [J]. NEURAL COMPUTATION, 2007, 19 (10) : 2756 - 2779
  • [22] Low Y., 2010, C UNC ART INT UAI
  • [23] Mackey L.W., 2010, P 27 INT C MACHINE L, P711
  • [24] Algorithms and Literate Programs for Weighted Low-Rank Approximation with Missing Data
    Markovsky, Ivan
    [J]. APPROXIMATION ALGORITHMS FOR COMPLEX SYSTEMS, 2011, 3 : 255 - 273
  • [25] Paterek A., 2007, P KDD CUP WORKSH4, P5, DOI [DOI 10.1145/1557019.1557072, DOI 10.1137/1.9781611972757.43]
  • [26] Salakhutdinov R., 2008, Proceedings of the International Conference on Machine Learning, P880, DOI 10.1145/1390156.1390267
  • [27] Schneider J., 2010, PROC SIAM INT C DATA, P211, DOI DOI 10.1137/1.9781611972801.19
  • [28] Scalable Coordinate Descent Approaches to Parallel Matrix Factorization for Recommender Systems
    Yu, Hsiang-Fu
    Hsieh, Cho-Jui
    Si, Si
    Dhillon, Inderjit
    [J]. 12TH IEEE INTERNATIONAL CONFERENCE ON DATA MINING (ICDM 2012), 2012, : 765 - 774
  • [29] Zhou YH, 2008, LECT NOTES COMPUT SC, V5034, P337, DOI 10.1007/978-3-540-68880-8_32