KBMF: Kernelized Bayesian Matrix Factorization [Download]

This software package implements kernelized matrix factorization with a fully Bayesian treatment and with an ability to work with multiple side information sources expressed as different kernels. Kernel functions have been introduced to matrix factorization to integrate side information about the rows and columns (e.g., objects and users in recommender systems), which is necessary for making out-of-matrix (i.e., cold start) predictions. Our algorithm has four main parts: (a) kernel-based nonlinear dimensionality reduction, (b) multiple kernel learning, (c) matrix factorization, and (d) binary classification (or regression).

If you use any of the algorithms implemented in the package, please cite one of the following papers:

• Mehmet Gönen and Samuel Kaski. Kernelized Bayesian matrix factorization. IEEE Transactions on Pattern Analysis and Machine Intelligence, 36:2047-2060, 2014. [manuscript]
• Mehmet Gönen, Suleiman A. Khan, and Samuel Kaski. Kernelized Bayesian matrix factorization. In Proceedings of ICML 2013, the 30th International Conference on Machine Learning, volume 28 of JMLR W&CP, pages 864-872, 2013. [manuscript]

This is experimental software provided as is; we welcome any comments and corrections but cannot give any guarantees about the code. If you have any comments or bug reports, please direct them to Mehmet Gönen.