/second-moment-loss

The second-moment loss (SML) is a novel training objective for dropout-based regression networks that yields improved uncertainty estimates.

Primary LanguageJupyter NotebookMIT LicenseMIT

Stargazers