I love how there are these two parallel threads going in HNville: (1) COLLEGE SUCKS BURN IT DOWN and (2) "I'd like to learn complex things like machine learning and math that take more than reading 5 blog posts to master, how do I do it"?

Sometimes this place just cracks me up.

I honestly think the answer is pretty simple: go to college. It doesn't have to be expensive. Take a community college course in calculus or undergraduate-level probability. Skip the gen eds and don't worry about the degree if you want to learn something narrow like this.

In any case, just find a mentor. On-the-job if you can, otherwise pay for a class.

What you shouldn't do is try to self-study by reading a book. You can perhaps do this but only if you're smarter than average and more motivated than most. Since you probably aren't, just take a class. Night school, maybe a MOOC. Preferably something heavy on analysis or proofs.

But you should do it with others. Math is a very social discipline, it's good to be able to discuss and have partners to work through things when you get stuck. And if you're like me, you WILL get stuck on things. This stuff is hard.

Another thought: this whole "college is great"/"college is terrible" dichotomy seems to occur people people don't think enough about quality. I think bad colleges are terrible and great ones are fantastic. I don't know any way I'd have learned all the complex topics in math, stats, probability, etc., I did without attending a big 10 engineering school (UIUC in my case)

I went to college and took a sum total of two math classes but have learned enough to reason out the many papers that don't have (pseudo) code but 'complex mathematical notation' -- which seems like almost all of them.

https://github.com/Jam3/math-as-code was helpful...