Join GitHub today
No description, website, or topics provided.
Fetching latest commit…
Cannot retrieve the latest commit at this time.
|Failed to load latest commit information.|
copyright (c) 2014, Yusuke Mukuta code was developed at Machine Intelligence Laboratory, the University of Tokyo. If you have any questions or suggestions, please contact us by email. Yusuke Mukuta firstname.lastname@example.org Introduction This program is an implementation of Bayesian Partial CCA and its variants. This program is developed for the following paper: Yusuke Mukuta, Tatsuya Harada, "Probabilistic Partial Canonical Correlation Analysis", In International Conference on Machine Learning, 2014. Please cite the paper if you use the code in a research. When I implemented the code, I refer to r code of CCAGFA (http://cran.r-project.org/web/packages/CCAGFA/index.html) for initialization. It depends on Optimization Toolbox for optimizing rotation matrices. Please use different library or eliminate this part if you don't have this toolbox. The calculation of inverse matrix may fail when the condition is too bad. In such a case, please add some regularization term. Please read demo_causal.m and demo_modelselection.m to learn how to use the code. The code empcca.m The implementation of probabilistic Partial CCA proposed in section 3.2. bpcca.m The implementation of Bayesian Partial CCA proposed in section 4.1. gspcca.m The implementation of Bayesian Partial CCA with isotropic noise proposed in section 4.2. demo_modelselection.m Sample code for model parameter estimation. Like the experiment in section 5.1. demo_causal.m Sample code for estimating causality measure. Like the experiment in section 5.2. makepccadata.m The code for making synthesized data generated by the probabilistic model. maketimeseries.m The code for making synthesized time-series data generated by the auto-regression model. partialccafortime.m The implementation of Partial CCA for calculating causality measure. corrs.m The code for calculating correlation in Bayesian model. infer_gs.m infer_b.m infer_bic.m infer_cv.m The codes for model selection par each methods: Bayesian Partial CCA with isotropic noise, Bayesian Partial CCA, Probabilistic Partial CCA with Bayesian Information Criterion, Probabilistic Partial CCA with 5-fold Cross-Validation. Licence This software is released under the MIT Licence.