Thread by @mrdbourke

1/ Introducing the 2020 #machinelearning roadmap:

An interactive mindmap which connects many (not all) of the most important concepts in machine learning.

Map: dbourke.link/mlmap
Video walkthrough:
Accompanying slides: github.com/mrdbourke/mach…

2/ In the map you'll find 5 branches:

1. 🤔 Problems - some of the main use cases for ML.
2. ♻️ Process - what does a solution look like?
3. 🛠 Tools - how can you build your solution?
4. 🧮 Math - ML is applied mathematics, what kind?
5. 📚 Resources - where to learn the above.

3/ Although very colorful, at first glance, the map can be very intimidating.

So there's a video walkthrough to go along with it:

We start with a high level overview which answers questions like "what is machine learning good for?"

4/ Then we cover a comparison between classical programming and ML.

E.g. if you wanted to write a program to cook your Sicilian Grandmother's famous roast chicken, you might start with the ingredients and recipe.

ML looks at the ingredients and dish then figures out the recipe.

5/ After the high-level overview and discussing some of the main problems ML is great for, we go through the typical process in a machine learning project.

Specifically:

1. Data collection
2. Data preparation
3. Training a model
4. Analysis/evaluation
5. Serving
6. Retraining

6/ But generic examples of how things are done aren't very fun.

So how about an actual project?

Sure, there's that too.

Here's the process related to a project I did recently to replicate @Airbnb's computer vision powered amenity detection: dbourke.link/airbnbplaylist

7/ So you're ready to work on your own ML projects, what tools can you use?

Here's a handful.

Including @weights_biases for tracking experiments and @streamlit for building apps.

ML tools are evolving fast, for more on this, check this post by @chipro: huyenchip.com/2020/06/22/mlo…

8/ "I want to get into ML, do I need to know math?"

Yes.

ML is using math to find patterns in numbers.

If your high school teacher didn't impart a love for math on you, it's time to take it on a second date.

You'll fall in love when you realise math is the language of nature.

9/ "Woah... this is a lot to take in, where can I learn it?"

Good news. We cover that too.

The short answer: many places.

But how about we start with some books + the internet?

Here's an example beginner-friendly (with a Python code flavour) curriculum:

10/ "Has anyone actually learned this stuff themselves?"

Yes, I have (thanks to the amazing creations of others).

I created my own AI masters degree: dbourke.link/aimastersdegree

Jason Benn did too, check out his curriculum here (I prefer this one): jasonbenn.com/post/how-i-lea…

11/ Be sure to play around with the map.

There are *plenty* of linked resources for learning more so follow your own curiosity.

If you're an auditory learner, the (movie length) video walkthrough adds a little riff about each section with examples:

12/ Of course, there are things missing (many textbooks could be written about the field and still not cover everything).

And the disclaimer is: many of the concepts/resources are biased from my own experience.

If you think something should be added/changed, let me know.

13/ Finally, all of this wouldn't have been possible without the huuuuge work from:

- Daniel Formoso's original machine learning mindmaps.
- @aureliengeron's Hands-On ML text.
- A plethora of online resources (many linked in the roadmap).

Happy (machine) learning!