Online update coeffs. for regression with Laplace noise.
[loglik,K1,K2] = c_reg_lapl(likpar,y,mx,varx)
[loglik,K1,K2] = c_reg_lapl(likpar,y,mx,varx)
- returns the coefficients for
the online update of the GP regression with Gaussian noise assumption.
Parameters:
likpar
- likelihood parameters (stored in net
).
y
- observed noisy output.
mx
- mean of the GP marginal at the new input.
varx
- the variance of the GP marginal at the new input.
loglik
- the logarithm of the averaged likelihood function.
K1,K2
- the first and second derivatives of loglik
.
The noise parameter lambda=likpar
is set at the initialisation.
The likelihood function is
P(y|x,lambda) = exp(-abs(x-y)/lambda)/(2*lambda);
(see matlab code for details).
ogp
, ogptrain
, ogpfwd
, err_abs
Copyright (c) Lehel Csató (2001-2004)