You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, only GPs output point estimates + uncertainties/variances, but the latter can be important.
We can do two things to get models with uncertainties:
implement more models that natively output variances (neural processes, bayesian networks)
or estimate variances using frequentist methods (bootstrapping / parametric bootstrapping). This is non-standard in machine learning, but worth thinking about, as this could give us variances for all models.
The text was updated successfully, but these errors were encountered:
Currently, only GPs output point estimates + uncertainties/variances, but the latter can be important.
We can do two things to get models with uncertainties:
The text was updated successfully, but these errors were encountered: