Ok, so it seems like we’ve established that all we need are vanilla GANs and vanilla LSTMs with well-tuned hyperparameters. All that’s left are getting rid of all those CNN hacks now?
Maybe we can just solve every machine learning problem with linear regression if we spend a *lot* of time tuning the learning rate? 
-
-
Then we can also go back to numpy…
-
Then I (or we) can just go back to the home country...
End of conversation
New conversation -
-
-
This is where a quantum computer will be handy; just try all learning rates in one go
-
Or Borg. The next best thing to a quantum computer.
- Show replies
New conversation -
-
-
need one hidden layer for non-linear decision boundaries
-
Not with the right features :) And if you argue that this requires “hand-crafting” those features, how is that different from what you’re doing with those complex NN architectures?
- Show replies
New conversation -
-
-
Oui?
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
If we toss out those pesky incorrectly classified points as "outliers" then we can do perfectly on every task!
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.