Computes the mean-square error.
[mseAvg] = err_mse(likpar,y,mX,sigX2)
[mseAvg] = err_mse(likpar,y,mX,sigX2)
- returns the mean-square error
between the training outputs y
and the posterior mX
.
Called with two output arguments, the function returns the individual squared distances as a second output argument.
The input argument likpar
is not used.
Parameters:
net
- the Gaussian Process data structure.
y
- desired output.
mx
- mean of the GP marginal at the new input.
allErr
- the mean-square error.
Copyright (c) Lehel Csató (2001-2004)