Have you ever played Civ?
You start the game by choosing your historical figure, set up your gameplay, and are suddenly dropped in the middle of nowhere. Well, you’re somewhere - but everything on the map is obscured, completely shrouded in an impenetrable fog. Which in a lot of ways is what it feels like when you decide that this is it. This time you’re going to learn Data Science//Machine Learning//AI//Deep Learning//the “it” thing of the moment. And if you’re like me and my first attempts at both Civ and Machine Learning, it feels a lot like this:
Civ Map <> Learning Map
The first time I played Civ I thought the map was broken, so I quit and didn’t touch the game again for three years. What I didn’t realize at the time was that with each turn you play in Civ, you begin to progressively uncover more and more of the map, until you’ve uncovered everything and can get a full lay of the land. To me there’s no greater metaphor for your personal learning journey than uncovering the map in Civ. So many of us set out to become “data scientists” or “machine learning engineers” because they seem like really cool things to do (and they are). And while there might be some general guidance (folks recommending this book or this class), no one knows your personal learning map - it’s something that you have to uncover as you go.
One strategy is to spend your first couple hundred turns building nothing but Scouts and sending them out into the world until you completely uncover the map. On the one hand this seems somewhat reasonable, because after awhile you’ll know everything that’s out there and then you can start committing your resources to a specific win condition. This strategy isn’t unlike the procrastiplanning I’m guilty of, where rather than jumping in and trying to learn something, I spend weeks and even months charting out the perfect learning path.
“Once I have this perfect path I’ll be ready to learn! It’s foolproof! I AM A GENIUS!” I tell myself as I Google another “best way to learn Python for real this time” resource.
But this strategy can leave you vulnerable and exposed, and it’s an awful lot of time you’ve spent not learning. And honestly, there are only so many Venn diagrams you can look at.
My Landmarks: Deep Learning and Magenta’s NSynth Super
I’ve spent a lot of time uncovering my map. Like, a lot of time. And I don’t have a perfect learning plan - I don’t think I ever did. Instead I have a couple of general landmarks that I know I want to work towards coupled with a good understanding of where I’m starting.
What helped me to find my landmarks was encountering two resources that instantly cleared away some of the fog. The first was Janelle Shane’s “You Look Like A Thing And I Love You”, which explores the absurdities of what you can create with AI, and the second was NSynth Super project from Magenta. I learned about the NSynth Super from this Nat & Friends video, and as soon as the video finished I was hooked - I needed to learn how to do this.
But even when some of your landmarks are visible, it doesn’t mean that the path is clear, or that it isn’t scary, or that you’re entirely confident that you can reach your landmarks. It’s kind of like playing single-player mode in a video game that you know nothing about. You might have a lot of fun, but you might also be frustrated and confused and eventually hit a barrier that you don’t know how to overcome.
Play Multi-player When You Can
Your learning map is probably going to look different from mine, because you’re a different person with different goals and a different background! But that doesn’t mean that we’re both not on a learning journey together. Sometimes you’re going to have to learn on your own, but as often as possible, spend time learning with others.
I have many crises of confidence - sometimes I spend entire days near tears because my landmarks feel so far away, the path is overgrown and dark, and I simply don’t want to do this anymore. But I also have people I can reach out to, from friends who can empathize with me to online forums where I can ask for help. You have to be tenacious to learn. It won’t always be easy, but it will always be worth it.
Learn With Me!
Every week I’ll be posting a round-up of my (machine and deep learning) learning journey. I’ll be sharing:
- What I learned
- How I learned it (both the resources I used as well as learning techniques)
- What I’m wondering
- What I’m stuck on
- What I’m planning to work on in the upcoming week
Learn With Me!
I truly believe that we learn better when we learn together. I’d love to hear what your landmarks are and how your learning journey is going, and the best place to do that is to start the conversation on Twitter!