18th-century math technique shows simpler AI models don't need deep learning

techxplore.com

3 pointsdanboarder2 comments

Okay, so they've taken an old method and applied it to deep learning and they say it's better.

But I'm not seeing anything beyond that in this article.

I haven't read the linked paper yet, but I'm not seeing much meat in this article. Just giving us the abstract from the paper might have been better.

It's in the article (https://www.sciencedirect.com/science/article/pii/S092523122... ) but I think the posted article does make it more understandable than the paper/abstract title "Additive autoencoder for dimension estimation".