Keras has been reliably winning almost every deep learning competition on Kaggle. Here are some thoughts as to why: https://www.quora.com/Why-has-Keras-been-so-successful-lately-at-Kaggle-competitions/answer/Fran%C3%A7ois-Chollet …
-
I can't think of any that wouldn't have "ease of use" as a primary objective. Are we talking
#sklearn wrt Kaggle? Or which ones?
Sklearn's defining characteristic was precisely ease of use. It stood out. *Most* specialized software libs aren't trying to be easy to use
-
-
I don't think ease of use is the defining characteristic of sklearn. It's that it was streamlined and well-documented.
-
Yes, the famous "lego bricks" :) Either way, what conditions necessary AND sufficient? Interesting question, even beyond
#Keras and Kaggle. - Show replies
New conversation -
-
-
Other ML libraries have tried to appeal to begineers (Rattle, Orange) but sklearn tried to be conceptually simple.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
The advantage there was that is was easier to compose parts of the system, and anticipate methods (.fit, .transform, .predict)
-
Final point, using keras means I spend less time futzing with shape parameters is reason enough to prefer it to other libraries.
End of conversation
New conversation -
-
-
Easy to use? Sounds like Keras :D
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
We indeed put focus on ease of use (via API design, docs, and good defaults). It's harder than people think, and seldom valued by geeks.
-
I super value it. Thanks for all the good work :)
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.