I took Linear Algebra during my first time at college, back when I was studying to be a chemical engineer, this was probably in early 2010, so thirteen years ago now. I remember enjoying the course, the material was interesting, and I believe that I did well in it, but can’t recall exactly what my grade was. And that was it. I couldn’t imagine a scenario where I would ever use or see it again, especially after I changed degree programs from Chemical Engineering to English.
And then I started this course and was fortunate enough to be assigned the AI/ML Trading Bot project for my capstone. I immediately started brushing up on machine learning through various combinations of Udemy courses and Youtube videos and started to learn how it all worked. Imagine my surprise when, halfway through the introductory videos for Tensorflow something clicked and I realized this was all just an application of Linear Algebra. I think the instructor was explaining one-hot encoding when I realized that what he’d just described was more or less an Eigenvector from Linear Algebra. And all that a tensor was was either a matrix, vector, or scalar from linear algebra with just a touch more abstraction. All of the tensor operations that the instructor had been explaining were vector transformations and matrix multiplication. I realized that I’d done most of this already and had never expected to see it again.
To say that this was a transformative realization might be overstating things just a tad, but it was certainly enlightening and it really highlighted for me, for maybe the first time, what people mean when they say that there is a difference between a computer scientist and someone who is just a programmer.
The emphasis in this degree program is almost overwhelmingly on practical coding–basically building applications–with some brief forays into more math-heavy concepts. And it seems, anecdotally at least, that students tend to struggle the most in those brief forays. And while I do enjoy building applications and some of the more practical aspects of the field like you might expect to see on the job. This dive into machine learning and some of the math behind it has gotten me genuinely excited to pursue it further.
Leave a Reply