DNNs are inadmissible as part of any hybrid model. 1) They are supervised and use backprop. 2) They suffer from overfitting. 3) They cannot see an object they have not been trained on. 4) They don't see 3D. 5) They don't get the physics of things. I could go on.
@botminds: yes and I first advocated hybrid models around the same time, in 1992, in my dissertation work with Pinker.https://twitter.com/botminds/status/955348500119769088 …
-
-
- 4 more replies
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.