dugzzuli / jmf Goto Github PK

Thank you for downloading this MATLAB package for Joint Matrix Factorization. This package contains the following files.

This folder contains Matlab code related to the following paper:

Lihua Zhang, Shihua Zhang. A Unified Joint Matrix Factorization Framework for Data Integration. Preprint arXiv:1707.08183. 
(https://arxiv.org/abs/1707.08183)

The folder was last updated by Lihua Zhang on July 31, 2017.

Please look `main_demo.m' at first and try to execute examples to learn how to use this software. Please send bug reports, comments, or questions to [email protected].

=================================
DESCRIPTION
=================================

The codes including three part: simulation part, main function part, prediction part
%%%%%%% simulation part
JMF_synthetic_dataset1.m
JMF_synthetic_dataset2.m
JMF_synthetic_dataset3.m
JMF_synthetic_dataset4.m
These four codes used for generating synthetic datasets.

%%%%%%% mian function part
The main function is named JMF.m, which includes many parameters used to select update rules and stop criterions. More details are illustrated in the function.
MUR_updateW.m, MUR_updateH.m
PG_updateW.m, PG_updateH.m
Ne_updateW.m, Ne_updateH.m
PANLS_updateW.m, PANLS_updateH.m
There codes used for updating W and H, which are invoked by JMF.m, JMF_prediction_L.m, and JMF_prediction_R.m.

%%%%%%% prediction part
Two prediction codes are included in this part, which are named by JMF_prediction_L.m and JMF_prediction_R.m.

%%%%%%% Other codes (compute_diff.m, compute_sumHRt.m, compute_sumtheta_record.m, compute_XHt_HHt.m, get_gradW.m, get_gradH.m, StopCriterion_rule1.m, and StopCriterion_rule2.m) are invoked by many codes mentioned above.