Machine Learning - My Learning Plan (Part 1)

My Machine Learning Journey

Hi! This is the beginning of my machine learning journey with you. I’ll dive into each concept, understand it deeply, apply it, and share my insights along the way.

I’ll be learning from Andrew Ng’s Stanford lectures, the famous StatQuest YouTube channel, and my favorite resource—Machine Learning Mastery by Jason Brownlee, PhD (a machine learning specialist and author of many ML books).

But before jumping into machine learning, there are a few things I need to prepare—starting with my mindset.

Overcoming Self-Limiting Beliefs

I used to hold myself back with thoughts like:

• What if I fail?

• What if I’m not good enough?

But the truth is—I AM good enough, and I’m capable of learning anything, no matter how hard it seems. Our brains are like muscles—they adapt and grow with practice.

Another mental block I had was thinking, "I can’t start until I have all the prerequisite knowledge." But often, these "requirements" are either unnecessary or so vast that waiting for them just delays progress.

I also used to make excuses like:

• "I don’t have a fast CPU/GPU."

• "I’m not a good programmer."

But here’s the reality: You don’t need the best hardware to begin, and programming skills improve with practice. Progress comes from consistent, step-by-step effort—and yes, some sacrifices, like all good things in life.

No more excuses! Let’s dive into why machine learning is such a cool field.

The Best Way to Learn Machine Learning? Top-Down Beats Bottom-Up

The Problem With Traditional (Bottom-Up) Learning

Universities love the bottom-up approach—structured, step-by-step, theory-first. Week by week, you crawl through concepts in a "logical" order. Sounds neat, right?

But here’s the catch: This isn’t how humans learn best.

Especially in machine learning, where you need to do more than just understand. As Jason Brownlee says:

"We are not robots executing a learning program. We are emotional humans that need motivation, interest, attention, encouragement, and results."

A Better Way: Top-Down Learning

Forget definitions. Forget "prerequisites." Start by connecting the subject to a result.

This isn’t just some fluffy advice—it’s how you actually learn things in real life:

• You learned to read by reading, not studying grammar.

• You learned to code by coding, not reading 500 pages on computability theory.

So why should ML be any different?

Why Top-Down Works for Me (And Probably You Too)

1. It’s iterative – You don’t "master" a concept in one go. You circle back, deeper each time.

2. It’s imperfect – Your first models will suck. That’s good. Sucking is step one to not sucking.

3. It’s yours – No syllabus. No arbitrary order. You follow what matters to you.

Most importantly: You’re not training to be a researcher. You’re learning to use tools to solve problems. That means you don’t need to swallow every equation before writing your first line of code.

The Bottom Line

Theory has its place—later. First, build something. Break it. Fix it. Repeat. That’s how real learning happens.

---

What’s your learning style? Bottom-up or top-down? Let me know in the comments!