Yeah. Serious overfitting.
The most surprising thing from NIPS so far is the continued prominence of MNIST in experiments.
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
for a while investigating issues in MNIST as a dataset was a hobby of mine. perhaps i should collate thosehttps://twitter.com/dribnet/status/696240671498240002 …
-
I've done similar investigations for Mikolov PTB in LMs ;) Also, re: MNIST,
@benhamner made this:https://twitter.com/benhamner/status/805864969065689088 …
- Show replies
New conversation -
-
-
Just like the prevalence of drosophila in genetic research- yet we don't see overfitting there (one hopes!)
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
appreciation where it is due I guess.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Yeah, I noticed that too. I mean even ImageNet is almost saturated these days, yet people choose to go with MNIST
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I feel like someone should make a short documentary on the history of the MNIST dataset and it's impact etc.
@jackclarkSF maybe?Thanks. Twitter will use this to make your timeline better. UndoUndo
-
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.