Bayesian Matrix Factorization with Non-Random Missing Data using Informative Gaussian Process Priors and Soft Evidences

Bence Bolgár, Péter Antal
Proceedings of the Eighth International Conference on Probabilistic Graphical Models, PMLR 52:25-36, 2016.

Abstract

We propose an extended Bayesian matrix factorization method, which can incorporate multiple sources of side information, combine multiple \empha priori estimates for the missing data and integrates a flexible missing not at random submodel. The model is formalized as probabilistic graphical model and a corresponding Gibbs sampling scheme is derived to perform unrestricted inference. We discuss the application of the method for completing drug–target interaction matrices, also discussing specialties in this domain. Using real-world drug–target interaction data, the performance of the method is compared against both a general Bayesian matrix factorization method and a specific one developed for drug–target interaction prediction. Results demonstrate the advantages of the extended model.

Cite this Paper


BibTeX
@InProceedings{pmlr-v52-bolgar16, title = {{B}ayesian Matrix Factorization with Non-Random Missing Data using Informative {G}aussian Process Priors and Soft Evidences}, author = {Bolgár, Bence and Antal, Péter}, booktitle = {Proceedings of the Eighth International Conference on Probabilistic Graphical Models}, pages = {25--36}, year = {2016}, editor = {Antonucci, Alessandro and Corani, Giorgio and Campos}, Cassio Polpo}, volume = {52}, series = {Proceedings of Machine Learning Research}, address = {Lugano, Switzerland}, month = {06--09 Sep}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v52/bolgar16.pdf}, url = {https://proceedings.mlr.press/v52/bolgar16.html}, abstract = {We propose an extended Bayesian matrix factorization method, which can incorporate multiple sources of side information, combine multiple \empha priori estimates for the missing data and integrates a flexible missing not at random submodel. The model is formalized as probabilistic graphical model and a corresponding Gibbs sampling scheme is derived to perform unrestricted inference. We discuss the application of the method for completing drug–target interaction matrices, also discussing specialties in this domain. Using real-world drug–target interaction data, the performance of the method is compared against both a general Bayesian matrix factorization method and a specific one developed for drug–target interaction prediction. Results demonstrate the advantages of the extended model.} }
Endnote
%0 Conference Paper %T Bayesian Matrix Factorization with Non-Random Missing Data using Informative Gaussian Process Priors and Soft Evidences %A Bence Bolgár %A Péter Antal %B Proceedings of the Eighth International Conference on Probabilistic Graphical Models %C Proceedings of Machine Learning Research %D 2016 %E Alessandro Antonucci %E Giorgio Corani %E Cassio Polpo Campos} %F pmlr-v52-bolgar16 %I PMLR %P 25--36 %U https://proceedings.mlr.press/v52/bolgar16.html %V 52 %X We propose an extended Bayesian matrix factorization method, which can incorporate multiple sources of side information, combine multiple \empha priori estimates for the missing data and integrates a flexible missing not at random submodel. The model is formalized as probabilistic graphical model and a corresponding Gibbs sampling scheme is derived to perform unrestricted inference. We discuss the application of the method for completing drug–target interaction matrices, also discussing specialties in this domain. Using real-world drug–target interaction data, the performance of the method is compared against both a general Bayesian matrix factorization method and a specific one developed for drug–target interaction prediction. Results demonstrate the advantages of the extended model.
RIS
TY - CPAPER TI - Bayesian Matrix Factorization with Non-Random Missing Data using Informative Gaussian Process Priors and Soft Evidences AU - Bence Bolgár AU - Péter Antal BT - Proceedings of the Eighth International Conference on Probabilistic Graphical Models DA - 2016/08/15 ED - Alessandro Antonucci ED - Giorgio Corani ED - Cassio Polpo Campos} ID - pmlr-v52-bolgar16 PB - PMLR DP - Proceedings of Machine Learning Research VL - 52 SP - 25 EP - 36 L1 - http://proceedings.mlr.press/v52/bolgar16.pdf UR - https://proceedings.mlr.press/v52/bolgar16.html AB - We propose an extended Bayesian matrix factorization method, which can incorporate multiple sources of side information, combine multiple \empha priori estimates for the missing data and integrates a flexible missing not at random submodel. The model is formalized as probabilistic graphical model and a corresponding Gibbs sampling scheme is derived to perform unrestricted inference. We discuss the application of the method for completing drug–target interaction matrices, also discussing specialties in this domain. Using real-world drug–target interaction data, the performance of the method is compared against both a general Bayesian matrix factorization method and a specific one developed for drug–target interaction prediction. Results demonstrate the advantages of the extended model. ER -
APA
Bolgár, B. & Antal, P.. (2016). Bayesian Matrix Factorization with Non-Random Missing Data using Informative Gaussian Process Priors and Soft Evidences. Proceedings of the Eighth International Conference on Probabilistic Graphical Models, in Proceedings of Machine Learning Research 52:25-36 Available from https://proceedings.mlr.press/v52/bolgar16.html.

Related Material