Intuitively, differentiable layered representation learning (deep learning) is in some ways closer to classical signal processing than it is to other branches of AI and ML, due to its substrate (stack of learned linear filters with a hardcoded differentiable non-linearity)
-
-
Show this thread
-
I wonder how differently the field would have evolved if we had chosen to use terminology derived from the world of signal processing instead of neuropsychology
Show this thread
End of conversation
New conversation -
-
-
Check out
@LightOnIO, by@IgorCarron et al. -
Indeed, we use coherent light for Machine Learning. And soon enough, people will be able to use our technology. We will have our next newsletter soon with more details: http://eepurl.com/cv_1h5
End of conversation
New conversation -
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
On a related note
@alirahimi0 recently blogged on the things deep learning can adopt from optics.https://twitter.com/TheShubhanshu/status/956630915152785408 …Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
The emerging of deep learning in cavity quantum electrodynamics (?)
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Can you backpropagate the light?
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
@alirahimi0 has an interesting take on this question of analogy with optics recently as well http://www.argmin.net/2018/01/25/optics/ …Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
'convolutional filter bank', just like 'wavelet filter bank'. A wonderful idea.
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.