/sci/ - Science & Math » Thread #13662487

55KiB, 840x400, deep-learning-vs-machine-learning-readspeaker-840x400.jpg

View Same Google iqdb SauceNAO

Anonymous Mon 20 Sep 14:29:21 2021 No.13662487 View Reply Original Report

Quoted By: >>13662529 >>13662536 >>13662538

instead of focusing on deep learning, why don't we experiment with shallow learning?

Anonymous

Anonymous Mon 20 Sep 2021 15:04:51 No.13662529 Report

Quoted By:

>>13662487
Some papers have shown good results with wider (rather than deeper) networks in certain fields. The problem is that it's hard to know which functions require lots of nonlinearities (or however you want to think of the equivalent in transformers) ahead of time, and it's easier to assume they need more than it is to assume fewer.

Anonymous

Anonymous Mon 20 Sep 2021 15:08:56 No.13662536 Report

Quoted By:

>>13662487
You mean wide neutral networks instead of deep neutral networks?
We do

Anonymous

Anonymous Mon 20 Sep 2021 15:10:20 No.13662538 Report

Quoted By:

>>13662487
Because hierarchical structures are inherently more scalable.

Capcode	All Only User Posts Only Moderator Posts Only Admin Posts Only Developer Posts
Show Posts	All Only With Images Only Without Images
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Ghost Posts	All Only Ghost Posts Only Non-Ghost Posts
Post Type	All Only Sticky Threads Only Opening Posts Only Reply Posts
Results	All Grouped By Threads
Order	Latest Posts First Oldest Posts First

Your latest searches