Decoupled Neural Interfaces using Synthetic Gradients · HackerPapers discussions

Decoupled Neural Interfaces using Synthetic Gradients — mentioned by throwawaymaths

Thu, 18 May 2023 11:42:58 GMT

https://arxiv.org/abs/1608.05343

Paper and discussion history. A mention is not an endorsement.

Decoupled Neural Interfaces using Synthetic Gradients — mentioned by dnautics

Sat, 14 Dec 2019 17:38:58 GMT

They don't, that's my point. I can find only one library for this: https://arxiv.org/abs/1608.05343

Paper and discussion history. A mention is not an endorsement.

Decoupled Neural Interfaces using Synthetic Gradients — mentioned by ilaksh

Thu, 23 Mar 2017 11:55:19 GMT

"Wild overestimation of progress" has been part of the story for many technologies. Minsky was very optimistic, and then he got burned by the AI winter, and became very pessimistic about AI. There is quite a lot of very serious investment aimed at this type of 'real' AI now. For example Deep Mind, DARPA L2M, recent $100 billion Vision Fund that Softbank's founder said was intended to bring about the singularity. I believe it because I can basically see how to handle most of the challenges using embodiment or virtual embodiment and interactive learning/L2M etc., by combining or taking…

Paper and discussion history. A mention is not an endorsement.

Decoupled Neural Interfaces using Synthetic Gradients — HN submission

Thu, 29 Sep 2016 13:33:54 GMT

Decoupled Neural Interfaces Using Synthetic Gradients

1 point; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Decoupled Neural Interfaces using Synthetic Gradients — HN submission

Fri, 19 Aug 2016 04:26:43 GMT

Decoupled Neural Interfaces Using Synthetic Gradients

6 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.