Rats generalize better than deep RL, and are more sample efficienthttps://twitter.com/IFLScience/status/1187008909246115841 …
1:00
Show this thread
To make this work at all, you have to carefully tune your learning rat
-
-
They’ve already been ‘tuned’ for millions of years to perform most of the components of this kind of activity. Hardly surprising.
-
This excuse is getting old.
- Show replies
New conversation -
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
"Rats have an innate prior thanks to millions of years of evolution"
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
This Tweet is unavailable.
-
It is gradient student descent anyway
End of conversation
-
-
-
What is the default learning rate for the Algernon optimizer?
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Choose your cheese decay carefully.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
But if you do that, will the rat generalize to other tasks?
-
What's the bias/variance tradeoff in car driving rats look like?
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.