I began researching machine learning in a serious way last December. Rather than a somewhat straightforward transition like applying my marketing background to an AI-powered product, I’ve been diving into the fundamental math that underlies machine learning.
In particular, I’ve been studying the excellent Deep Learning textbook by Ian Goodfellow, Yoshua Bengio, and Aaron Courville. In order to engage with the material, I’ve spent months re-learning and learning for the first time several topics including linear algebra, vector calculus, statistics, Bayesian probability, formal definitions of functions and vector transformations, and even some elementary manifolds.
I’m hardly alone in believing that machine learning is a powerful, fundamental shift in the tech industry. But I think people still don’t quite get it. My studies feel a bit like someone learning HTML, HTTP, and other Internet standards circa 1994. If your risk-averse parents asked why you weren’t studying for the LSAT, you could hardly tell your mom and dad that you were planning on being an Internet entrpreneur in 1998 or a professional manager of SEM in 2005. But I’m convinced that 5 and 10 years down the line, my homework now is going to pay dividends.