I see no evidence at all that this is better. Not saying it isn't, just saying I see no evidence...https://twitter.com/lukede0/status/901193689938001920 …
Looking at the images, it seems to suffer from the same fundamental issue as MNIST: no translation invariance (unlike CIFAR10)
-
-
Fair points, though for a baseline this could be slightly better than comparing out to the thousands digit on accuracy.
-
The page doesn't try any modern approaches on the dataset, so it's not clear whether it's easy, or whether insights will generalize
- Show replies
New conversation -
-
-
Perhaps an uncentered version of this would've been better
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.