covariance computation with loss hessian

chariff / GPro

Python package for Preference Learning with Gaussian Processes.

MIT License

31 stars 7 forks source link

Closed KrisNguyen135 closed 3 years ago

KrisNguyen135 commented 3 years ago

A Laplace instance returns the negative log likelihood Hessian in addition to the optimized mean posterior.
The predict() method in ProbitPreferenceGP takes in an additional optimal argument to return the covariance matrix.

chariff commented 3 years ago

Thank you very much.