Performs a full online update step of the Gaussian Process
ogpstep_full(cX,indX,kX,kk,K1,K2,gamma,hatE);
[net] = ogpstep_full(net,cX,indX,kX,kk,K1,K2,gamma,hatE)
- updates the
GLOBAL Gaussian process structure net
with the information from the
current data point. This function is called from ogppost
and this
function makes no change inthe EP-representation of the GP (the chenges are
done in ogpstep_ep
).
The BV set is extended with cX
- the current input example.
gamma
and hatE
are the geometric components required to update
the inverse of the Gram, or kernel matrix.
net
- the GLOBAL Gaussian Process structure to be modified.
cX
- the current input that will be included into the basis vector set.
indx
- the index of the current input in the training set (needed to update the TAP/EP representation).
kX
- the covariance function of the basis vector set and the current input K0(BV,x).
kk
- the variance at the current input K(x,x)
.
K1,K2
- the values of the first and second derivatives of the log-average.
gamma
- the geometric component of the updates (the distance of the current input from the subspace given by the basis vector set.
hatE
- the coordinates of the projection of the current input to the basis vector set.
ogp
, ogptrain
, ogpstep_sp
, ogpstep_ep
Copyright (c) Lehel Csató (2001-2004)