Mike Grosskopf

Results 2 comments of Mike Grosskopf

It appears your correlation length is very small. Did your loss converge? If so, you may have just converged to a local minimum. It's good practice with GPs to run...

@deng-cy Did you randomly initialize the starting parameters for the optimization? If you started from the same spot, L-BFGS will always take you to the same minimum. Another possible issue:...