How I learnt Machine Learning

About me

I am a software engineer by training and profession.

Why I wanted to learn Machine Learning

I was motivated to learn machine learning while I was in college. I had interest in statistics and since I was being trained to code performant computing systems, I decided to learn something that lies at the intersection of these two.

How I approached learning Machine Learning

I started with Stanford Machine Learning course(cs229) by Andrew Ng. After that, I read Pattern Recognition and Machine Learning by Christopher M. Bishop. Then whenever required I read books/articles on prerequisites like optimization techniques, estimation theory, etc.

Challenges I faced

There were times when to understand a derivation, I had to read multiple topics, sometimes whole book because I didn't have the background knowledge necessary. This was frustrating, but it also helped me gain in-depth knowledge of what I was studying.

Key takeaways

I gained knowledge in diverse but related fields. Like I never knew how I would use topology in Machine Learning. But, turns out it is important to prove certain results directly or indirectly.

Tips and advice

Learn the maths behind algorithms instead of just using it on some dataset. Go deeper into the results. It is time consuming, but it will pay off.

Final thoughts and next steps

Stats, Linear Algebra, Optimization. Get a good hold on these topics before starting machine learning. It would really easy and smooth that way.