Re:Predictive coding has been unified with backpropagation
by tsmithe on Monday April 05, 2021 @02:42PM
(#26702865)
In fact, the compositional structure underlying that of predictive coding [0,1] is abstractly the same as that underlying backprop [2]. (Disclaimer: [0,1] are my own papers; I'm working on a more precise and extensive version of [1] right now!)
[0] https://arxiv.org/abs/2006.01631 [1] https://arxiv.org/abs/2101.10483 [2] https://arxiv.org/abs/1711.10455
Re:Predictive coding has been unified with backpropagation
by eli_gottlieb on Monday April 05, 2021 @04:02PM
(#26703892)
Hurry and publish before I have manuscripts ready applying these results.
Re:Predictive coding has been unified with backpropagation
by tsmithe on Monday April 05, 2021 @04:25PM
(#26704210)
Hey, Eli :-)
I'm working on it; I'll send you an e-mail. Things quickly turned out to be more general than I realized last year.
Re:Predictive coding has been unified with backpropagation
by eli_gottlieb on Monday April 05, 2021 @08:14PM
(#26706406)
That would make sense. The whole ACT/categorical cybernetics community has been working out how massively general optics are :-).
Re:Predictive coding has been unified with backpropagation
by selimthegrim on Monday April 05, 2021 @03:28PM
(#26703477)
What were you going to say about Young tableaux?
Re:Predictive coding has been unified with backpropagation
by adamnemecek on Monday April 05, 2021 @03:36PM
(#26703578)
Dynamic programming and reinforcement learning are just diagonalizations of the Young tableau. This is related to the spectral theorem.
Re:Predictive coding has been unified with backpropagation
by selimthegrim on Tuesday April 06, 2021 @04:16PM
(#26716835)
Interesting. Is there some concrete realization you had in mind?
I wrote something about this here https://github.com/adamnemecek/adjoint