ML is difficult to understand without a technical background, but the fundamental ideas are accessible. Most introductions unfortunately either go into unnecessary details (e.g. the internals of neural networks), or, go to the other extreme and just provide vague wishy-washy analogies.
Hence, I wrote my own explainer, split into three parts:
- What is Orbit? (3-5 minutes): I use Andy Matuschak’s Orbit system to make the posts more engaging. It’s not complicated, but without an intro you’d likely get confused.
- What is a function? (5-10 minutes): Almost all technical people take for granted that everybody knows what a function is, but it’s crucial to understand ML.
- A toy example of ML (15-25 minutes): This post walks through a concrete example of ML, highlighting the key ideas, e.g. what precisely is being learned during training.