It’s nearly time to let the machines loose.
With three classes left, we’ve finished learning about “supervised” machine learning problems, in which you know what sort of answer you want back: How expensive is a house, is a tumor malignant, is this series of pixels a 3 or an 8? Next, we’re on to “unsupervised” problems, where you feed the machine a bunch of data and trust it to find what sort of patterns exist. I’ll know more about this next week, after we’ve learned about it.
With the guts of machine learning mathematics behind us—dear God, I hope—it’s a convenient time to step back and consider whether this class has been a success so far. I chose the advanced track, in which students have to complete programming assignments in addition to review questions, because I wanted to pick up enough coding skills to possibly use this stuff in the future. (Last week, I mentioned it might be useful for creating a robot blogger.) But when it’s 4 a.m. and I’m struggling with the homework, I will confess that I have wondered: What is the point of an exercise like Stanford’s? Those of us taking this class remotely don’t get any sort of academic credit for completing it. It’s useful to have on a résumé, but I’m not sure how seriously employers will take this compared to a traditional, accredited online course. I’m looking forward to pinning my certificate of completion on my refrigerator, right next to the magnet with the number for Domino’s on it. But is it that sufficient motivation?
I submit that the value in a course like this is that college-style learning is often wasted on those physically in college, and it’s a wonderful opportunity to continue exploring new areas of knowledge. I was lamenting with a friend the other day about how little I explored the thousands of fascinating classes when I was actually in college, getting real credit for all this work. But in many ways, the formative years of 18 to 22 are the time when many of us are least primed to dutifully attend lectures—particularly at ungodly hours like 10 a.m.—and immerse ourselves in a life of the mind. Distracted by parties and by the 60 hours a week I spent at the Cavalier Daily, I never would have had the wherewithal to take a math-intensive course like this as an undergraduate. Now that I have a little more appreciation for the infinite realms of knowledge yet to be explored, a class like this (for free, I should add) is a gift.
And I do hope that employers learn to value a course like this as much as one with some sort of credit attached. We’re not talking about watching a TED lecture here. These are graded classes with difficult homework and penalties for slacking off. Whether or not it gets me closer to a master’s in computer science seems irrelevant.
My only concern is a practical matter: that there is no good way to take this class without watching the video lectures in order. While the lecture notes are also posted online, they don’t make a tremendous amount of sense if you haven’t watched professor Andrew Ng mark them up with his stylus in real time. Of course, watching the videos is one major appeal of a class like this, because it creates a connection to the instructor—in this case, a wonderfully talented teacher—and simulates a classroom aesthetic from across the country. But one of the great challenges of teaching is that everyone learns differently, and I’m not someone who primarily absorbs information aurally. I enjoy watching the lectures, but to do the work I need to be able to go back to the lecture notes and study the equations.
Beyond that, I consider this experiment in distributed learning a great success so far for anyone who feels sufficiently motivated to participate. And of course, there’s always the advantage of doing the work on your own schedule—even if it creates a certain hazard that your studying will be interrupted when you light the sleeve of your bathrobe on the beeswax candle on your desk while doing the work. Which just happened.
