Hacker News

4 years ago by pishpash

Also, most recent state of progress: Predictive Coding Can Do Exact Backpropagation on Any Neural Network (2021)

https://arxiv.org/abs/2103.04689

4 years ago by obstbraende

here's a link with reviewer comments https://openreview.net/forum?id=PdauS7wZBfC (praise to openreview!)

4 years ago by pishpash

The decision reasoning is super helpful to put things in context. The arbitrary binary decision to "accept" vs "reject" especially for the snooty "high bar for acceptance at ICLR" is laughable in a world of free information access.

4 years ago by yewenjie

AstralCodexTen (formerly SlateStarCodex) has discussed this here - https://astralcodexten.substack.com/p/link-unifying-predicti...

He mostly points to this post in LessWrong - https://www.lesswrong.com/posts/JZZENevaLzLLeC3zn/predictive...

4 years ago by amelius

If backprop is not needed, would this finding make automatic-differentiation functionality obsolete in DL frameworks, allowing these frameworks to become much simpler? Or is there still some constant factor that makes backprop favorable?

4 years ago by lumost

Reading the openreview link, the current understanding is that this approach is dramatically more computationally intensive than standard backprop - limiting its utility.

4 years ago by p1esk

Backprop is simpler.

4 years ago by coolness

Needs [2020] in the title. Interesting work nevertheless.

4 years ago by l33tman

This (and its follow up papers) have already been discussed multiple times here. Don't have the links handy though..

Daily digest email

Get a daily email with the the top stories from Hacker News. No spam, unsubscribe at any time.