More details about Top-Down vs Bottoms-Up

[paranoiks]

I read the readme file, but I don't understand what is meant by going Top-Down vs Bottom-Up in this specific context. I am interested in hands-on experience from the beginning, delving into maths where necessary. Where should I start?

[zalimitless]

I believe the concept is as follows:

Top-Down

When you take the general knowledge of a topic and use it to learn how it could be used in applications, then making your way down to specific topics, such as the basics of math.

So you will first learn how everything is working on the high-level and how to quickly implement the applications.

Bottom-Up

Learning all of the fundamentals first, like mathematical concepts that will assist in developing your own tools for AI. Then working your way up to how everything is working together and finally how you can implement the applications.

Benefits of either/or

When learning Top-Down you can quickly learn how to implement AI in your applications, without fully understanding what is going on under the hood. This is great if you quickly want to have a solution to a problem that you are facing. But eventually you will learn about what is going on under the hood as you make your way down to the basics.

When learning Bottom-Up, you will fully understand the underlying logic and implementations on how AI works. This will allow you to be more flexible with your code, whereas you might even write your own AI in your applications or new innovations that could push the AI front.

Now, both have the same goal: Learn and Use AI.
It's just a preference: Do you want to know how to use it first? Or do you want to understand why it works first?

[zalimitless]

Anyways, this is just my own interpretation on your question.

Each approach can be quite simple—the top-down approach goes from the general to the specific, and the bottom-up approach begins at the specific and moves to the general. These methods are possible approaches for a wide range of endeavors, such as goal setting, budgeting, and forecasting.

Source: Investopedia

[paranoiks]

Thank you for the explanation, I understand that bit. My question was more along the lines of "which of those resources should I follow if I want to go top-down?"

[zalimitless]

Good question.
I imagine that the "Mathematical Foundations" would be the "Bottom" and the other "* Learning" topics would be the "Top".

So start with the "Machine Learning" topic, then make your way down to: "Deep Learning", "Reinforcement Learning", "Natural Language Processing" and then to "Mathematical Foundations"