Usual reminder: when I've been saying for the past 5+ years that deep learning is interpolative, I don't mean it does linear interpolation in the original encoding space (which would be useless). It does interpolation on a low-dimensional manifold embedded in the encoding space.
-
-
I see your point. But what about symmetry breaking? Would a lower dimension manifold dictate equal probability to spork/fooon at t=0.5 ot would it be biased towards real world sporks probably found mislabeled in the dataset?pic.twitter.com/qUjqCBFykT
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Yes, if numbers are just pixels on 2d grid. No, if numbers are multiple levels of abstraction of multimodal inputs (vision, tactile, manipulation, usage, hunger, taste, language, social...)
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.