Sparse Gaussian Process Regression Model (SGPR) with optimization through Variational Free Energy (VFE).
More...
#include <FgVFESparseGPR.h>
Private Attributes | |
double | dAlpha |
Sparse Gaussian Process Regression Model (SGPR) with optimization through Variational Free Energy (VFE).
VFE for SGPRs integrates out the input variables and computes/maximizes a lower bound on the exact marginal. This avoids overfitting and enables approximations for not tractable integrals in deeper GP models.
For more information see, http://proceedings.mlr.press/v9/titsias10a/titsias10a.pdf https://www.repository.cam.ac.uk/handle/1810/273833
, 01.04.2019.
Definition at line 41 of file FgVFESparseGPR.h.
|
private |
Definition at line 118 of file FgVFESparseGPR.h.