About SARM. At this point I am 100% convinced that the VGG16 experiment is not for real. Most likely a big experimental mistake, not fraud.
-
I have tried this exact same setup last year, explored every possible variant of that algorithm. I know for a fact that it doesn't work.
Better: I know why it doesn't work. Which isn't addressed in the paper.
-
-
Conditional on this interpretation of the algorithm used, there is a 0% chance that performance would be anywhere near backprop for VGG.
-
I cannot speak for the first two experiments. As for the VGG16 experiment, I am confident enough about this to state it publicly.
- Show replies
New conversation -
-
-
Care to explain why it doesn't work?
-
For context, this is paper we are talking about : https://arxiv.org/pdf/1608.04062v1.pdf …
End of conversation
New conversation -
-
-
Presumably the PCA version just gives you pixelwise-decorrelated feature maps, which might be ~useful but probably not that useful
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
about using features learned with PCA in the convolutational layers? Why would that not work? Makes sense from theoretical point.
-
Just curious as I never fiddled there like you did. BTW! It is good to know that does not 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.