Improving the Way I Learn

Yesterday was a slow day. I remember waking up close to 9.30, cooked a bad-tasting musli for breakfast, finished an assignment on Neural Style Transfer for my Convolutional Neural Networks course on Coursera, made eggs for lunch, watched a video where Elon Musk talks more about the Boring Company and tunnel-digging, then went off to watch more videos related to that topic. The video about the Boring Company and its technology was an inspiration, and I wanted to learn about how Elon Musk is able to apparently master so many different disciplines. Here's Elon Musk replying to a Reddit AMA about how he learns:

it is important to view knowledge as sort of a semantic tree -- make sure you understand the fundamental principles, ie the trunk and big branches, before you get into the leaves/details or there is nothing for them to hang on to.

I noticed that I've been starting to adopt a similar style as of a few days ago. While I am studying a topic on Coursera, I'd start by creating an empty sheet for the topic, starting with a question (e.g. what is neural style transfer?), answering that, and in the process more questions come up, and I answer those, etc. I was essentially creating a tree, as Elon Musk was suggesting.

As an example, here's a snippet I wrote while studying Neural Style Transfer:

Mine was top-down as opposed to bottom-up. I'd like to fuse both approaches together.

Later in the day, I decided that I wanted to read more about this topic, and then I landed on a book called "Make it Stick: The Science of Successful Learning". I didn't end up reading the book, but viewed a summary of it online. The concepts in this book were complimentary to my earlier thoughts:

  • People often get confused between fluency and mastery. I can read about a topic several times, but when I keep reading, I am just remembering words and lingo related to the topic. I get the illusion that I've mastered it just because the words seem familiar, but in reality I've just become "fluent" in it.
    • In some way, I've done this for big parts of my deep learning specialization courses online and I want to change that.
  • Retrieval is a far more effective technique than rereading. It feels harder, but that's to some extent the point. Challenging yourself in retrieving the content you learned, by testing yourself or writing about it, greatly accelerates the learning.
    • I feel this myself consistently. I remember a few years ago I joined a learnathon where I read and made a five minute presentation about how the GPS works. And, yes, I can still do a similar presentation today as I did then. Going through that process made the concept of a GPS really stick in my memory.
  • Interleaved learning is more effective than sequential learning. Say I'm learning soccer, I can practice by playing some penalty kicks, followed by a few corner kicks, etc. The author here argues that this isn't as effective as interleaving these activities together. When I repeat a micro skill over and over, I get the illusion that I am mastering it, but in reality I am relying a lot on short-term memory. It's more fluency than mastery.
    • I am not sure I fully agree with the author on this concept actually, but will let it sink in more.
  • Growth mindset is key. Believing that you can has a psychological boost to help you learn the topic at hand, and that's certainly something I felt first-hand.

Leave a Reply

Your email address will not be published. Required fields are marked *