Armchair linguist and self-taught coder battling with math phobia as I learn the numerical nuts and bolts of machine learning. The site’s name is an anagram of my name—nothing more.
Though modular arithmetic is rather obscure, we actually use it every day—primarily when we tell time. And if you’re a programmer, you have probably used the % operator before. However, there is much more to modular arithmetic, and the way it seems to turn normal math on its head is very fascinating.
The AKS test is an award-winning algorithm for proving whether an integer is prime, though I realized too late that it is too slow to be of any practical use. Writing the code itself was useful, though.
After my previous fail trying to use autograd with pandas, I decided to set simple linear regression aside and attempt training a neural network. Here, I’ve adapted the network in the Welch Labs tutorial to accommodate an arbitrary number of layers. (However, floating-point arithmetic doesn’t follow suit quite so easily!)
I got halfway through this post only to be thwarted by the limits of the autograd module. The issue I opened on their GitHub repo has so far fallen on deaf ears, so I lost interest in linear regression for the time being.