The paper is difficult to follow, so at first I wasn't sure about algo details. After some discussions, in particular with Kevin Murphy,
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.
-
-
-
it seems that the VGG16 experiment merely consists in using convolutional kernels obtained by computing the PCA/LDA of the input filter maps
- Show replies
New conversation -
-
-
there might be an important technical detail that you missed, eg the "skip connections" or in the way pca/lda is used to get F.
-
skip connections are absent in the VGG experiment (0th iteration approximation = relu).
- Show replies
New conversation -
-
-
I tried some cheating setting and got results similar to that claimed in the paper.Really worried that this is a sign of big fraud
-
what's your implementation? What's the "cheating settings"?
- Show replies
New conversation -
-
-
Epilogue: the manuscript was withdrawn by the first author. It looks like it may have been deliberate fraud.https://arxiv.org/abs/1608.04062
-
Wow. Looks like you called it (not fraud specifically, but invalidity of the results)
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.