File Exchange

image thumbnail

Fast multi-output relevance vector regression

version (10.3 KB) by Youngmin Ha
Multi-output nonlinear regression


Updated 18 Apr 2017

View License

This is a faster program than the the existing program uploaded at You can download the corresponding paper at

Cite As

Youngmin Ha (2020). Fast multi-output relevance vector regression (, MATLAB Central File Exchange. Retrieved .

Comments and Ratings (7)

Youngmin Ha

Yes, Monte Carlo simulation can be parallelized by using parfor.

Sione Palu

Instead of the while loop, can the code be modified to run parallel using the Parallel Toolbox?

Youngmin Ha

You can download my paper at

Niu jia

the paper link is available, but the download does not work,pls renew a link,thx

songtebo tebo

Youngmin Ha

You may use "Gauss" instead of "+Gauss" and delete "assert(size(Phi,2) == N + 1" if all regression outputs do not have any constant (i.e. bias). You should use "+" if you do not have any ideas about regression outputs.


I have been enjoying your code, I have 2 questions

1. Did you intend for a user to be able to unbiased the Kernel though specifying 'Gauss' instead of '+Gauss'? Removing the '+' sign causes an error in fmrvr on the line

assert(size(Phi,2) == N + 1, 'unexpected matrix size of Phi')

2. In fmrvr there is referenced your paper: "Fast multivariate relevance vector regression," to Annals of Mathematics and Artificial Intelligence (2015). Can you send me a link to this paper as I cannot locate it via searching online.


URL of corresponding paper has been added.

Title has changed as "multi-output" is more proper than "multivariate".

Thumbnail image is changed.

Old license is deleted.

MATLAB Release Compatibility
Created with R2013b
Compatible with any release
Platform Compatibility
Windows macOS Linux