2020 Kaggle survey: regression, decision trees and ensembles dominate in ML applications.pic.twitter.com/dZTzI7wskV
-
It seems all the time I invested in understanding kernel machines won't pay
I'm not surprised to see this however. Although MNIST is usually the "go-to toy dataset" when introducing neural nets, in real world scenerio this task is still tackled with logistic regression.
Kernel machines are still one of the most widely used and best ML methods. I don't know why they're not included in the Kaggle survey.
-
-
They have fallen out of use due to being largely very slow on CPU for even moderately sized datasets. I am hoping that changes now that super fast GPU accelerated options are becoming readily available.
-
- Näytä vastaukset
Uusi keskustelu -
Lataaminen näyttää kestävän hetken.
Twitter saattaa olla ruuhkautunut tai ongelma on muuten hetkellinen. Yritä uudelleen tai käy Twitterin tilasivulla saadaksesi lisätietoja.