I found my topic. "ForestNet: Towards an alternative to DeepForest" https://arxiv.org/abs/1702.08835 pic.twitter.com/rOD5VGGRDe
About the DeepForest paper itself: haven't read very carefully, but I don't understand what's new about it...
-
-
...it seems to be about a classic ensemble technique (cascading clf with input reinjection). This is old news, and not a competitor to DL.
-
Remember: stacking classifiers, even in a smart way, can never beat *joint training of a feature hierarchy*. "Joint" in particular is key.
End of conversation
New conversation -
-
-
I like how the paper title contains all the three choices from your poll. At least that's novel.
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.