Computes the log-predictive probability of the labels.
[logavg, logind] = err_2logp(likpar,y,mX,sigX2)
[logavg] = err_2logp(likpar,y,mX,sigX2)
- average log-predictive
likelihood for the data-set.
Parameters:
likpar
- the likelihood parameters.
y
- desired output.
mx
- mean of the GP marginal at the new input.
allErr
- the mean-square error.
[logavg, logind] = err_2logp(likpar,y,mX,sigX2)
- additionally to the
average, it also returns the individual errors.
demogp_class
, demogp_gui
, ogp
, c_class_bin
Copyright (c) Lehel Csató (2001-2004)