Explain each category?

[sradu]

This list looks amazing! One thing that I think would be cool for newbies is to explain the category in a paragraph or two.

For example, what is 'Architecture Search' and what is it useful for.

[ritchieng]

That sounds reasonable, I'll get to that soon.

[sradu]

Yay! Awesome!

[xen0f0n]

@ritchieng I'd be happy to write a few short descriptions for some categories.
Would it be ok to reference some popular papers, since a few lines are excerpts from those papers?

For example (on Multi-Task learning):

In Multi-Task Learning a model is trained on different related tasks. It has been observed that learning these tasks jointly can lead to performance improvements compared to learning them individually, taking advantage of the fact that related tasks may share informative features (Zhang, Qiang, 2017).
Having a single model perform different tasks simultaneously is also efficient in terms of training time and inference speed.
In computer vision, multi-task learning has been used for learning similar tasks such as image classification in multiple domains, pose estimation and action recognition, and dense prediction of depth, surface normals, and semantic classes (Liu et al., 2019).