How to use Gaussian Process Regression with multidimension input with meanfunc?

I have downloaded the Gaussian Processes for Machine Learning (GPML) package (gpml-matlab-v3.6-2015-07-07) from the website. The demo_regression for one dimension works just fine in the matlab.

Now I have my own data for regression where the xtrain (training data) is a 20*5 matrix (20 samples, 5 input vars), and the ytrain ( training target) is 20x1, test data xtest is (1x5), The problem is that I do not understand how to calculate the meanFunction, the code provided for the regression example does not work for multiple input datasets.

This is the error which appeared to me when I run the code:

..................................................................

Error using gp (line 90)
Number of mean function hyperparameters disagree with mean function

Error in minimize (line 75)
[f0 df0] = feval(f, X, varargin{:});          % get function value and gradient

Error in GPMLexample1 (line 108)
 Ncg =50;hyp = minimize(hyp0,'gp', -Ncg, inf, mean, cov, lik,TrainingSetX_x, TrainingSetY_x); % opt hypers

...................................................................

Are there anybody who has tried this, and maybe can show an example?

[Karim kim]

I've the same problem, please let me know if you figure out how to fix it.

Best regads

Karim

[somi]

you should use The ARD SE covariance function for multi-dimensional inputs.

Answer 1

[Don Mathis]

I suggest using the FITRGP function in the Statistics & Machine Learning Toolbox instead of GPML

Answer 2

[LUIS PEJENDINO]

what is meanfunc used for?