Supervised Learning: No Loss No Cry
Supervised learning requires the specification of a loss function to minimise. While the theory of admissible losses from both a computational and statistical perspective is well-developed, these offer a panoply of different choices. In practice, this choice is typically made in an \emph{ad hoc} man...
Guardado en:
| Autores principales: | , |
|---|---|
| Formato: | Artículo |
| Publicado: |
2020
|
| Materias: | |
| Acceso en línea: | https://arxiv.org/abs/2002.03555 http://arxiv.org/abs/2002.03555 |
| Etiquetas: |
Sin Etiquetas, Sea el primero en etiquetar este registro!
|