100 Days of ML Code: takeaways (part 1)

Siraj and his idea

Inspired by the challenge #100DaysOfMLCode, I pledged to dedicate 3 hours a day for the next 100 days to embrace the field of Machine Learning. This amount of hours is arbitrary. In fact, you can do less. Just do it every single day, without exceptions.

It’s been already one month since I started and here are 5 useful takeaways:

  1. Top-down or bottom-up approach — it doesn’t really matter. There is no such thing as ‘the best learning strategy’. The most important rule here — just learn! Don’t be afraid of black boxes — on the way there will be many of them. It’s totally fine! Every problem is solvable and if not — just pick another one.
  2. Find a learning resource that fits YOU the best. Don’t listen to anyone. There are plenty of platforms, books, online courses and YouTube-channels about Machine Learning, Deep Learning (and everything else). Don’t be too dramatic about your decision and just pick one learning resource. Stick to it as long as you see the progress. Once you notice it’s not enough anymore — choose another one. Or even use many of them simultaneously. My personal choice fell on the Deep Learning Specialization Course by deeplearning.ai , taught by Andrew Ng. During the first month of learning a chill step-by-step bottom-up approach worked for me perfectly. By the end of the first part of this Specialization you will be able to actually understand how the neural networks technically work and what math is essential for building them from scratch.
  3. If you are new to programming in general ( so am I ) , try to improve your coding skills on the way. If you don’t know what a function does — google it. If you don’t know how to push a repo to the GitHub — google it as well. Think of programming as a technical toolkit that turns sci-fi to reality. Since you’re blessed, this toolkit is available to you. So use it!
  4. Have clear goals in mind. Ask yourself: Do I really need to understand this concept in details? Or is it enough just to know its practical application? If you have a clear goal, it is easier to set priorities. Learn only what you really need and you won’t feel overwhelmed. Your decision making process will be much faster if you know what exactly you are looking for.
  5. And finally, take notes by hand. Really. And it’s not my personal choice. It’s been proven that it is more effective than laptop note taking. And of course way more effective than no notes at all.

That was some advice I wish I knew by the beginning of the challenge. Next month I will make another blog post about the second part of my journey. Please clap if it was helpful!

Source: Deep Learning on Medium