Live data from Hacker News

Understanding Machine Learning: From Theory to Algorithms

cs.huji.ac.il

51–56 of 56 posts

Re: Understanding Machine Learning: From Theory to Algorithms

#51
Is there utility for a software engineer to understand ML concepts to a deep degree if they don’t perform research?

Trying to gauge where I should focus learning for my career (which i don’t plan to do research in)

Roughly I see a gap in businesses needing AI/ML implemented, but outside some debugging, would it be worthwhile to develop a model from scratch or would some off the shelf model for use case X, tuned a bit, likely fit most use cases for a standard business?

Re: Understanding Machine Learning: From Theory to Algorithms

#52

What other books do people recommend?

An Introduction to Statistical Learning: https://www.statlearning.com

It is a free, easy introductory text by leading researchers that covers all the classics and includes many "Lab" sections with code.

It has a chapter on deep learning but doesn't cover any of the recent advances. You will need other sources for that.

Re: Understanding Machine Learning: From Theory to Algorithms

#54
post #47

Earlier quoted context omitted.

Most ML/AI roles have requirements for a strong mathematical background (at least what I have seen in germany). If you can show off some skills I still wouldnt completely rule it out. Reading a single book cover to cover wont cut it though imo.

If you have an undergraduate’s understanding of calculus and linear algebra, you’re as or more advanced than the legion of ML PhD candidates I see graduating all the time. A field like that is running on hype, and has no quality control at all. I’ve seen people get hired into Ivy League tenure track jobs without knowing how linear algebra really works.

I know ML PhD candidates with a very strong understanding of mathematics and statistic, but it might depend on where you are.

Re: Understanding Machine Learning: From Theory to Algorithms

#55
post #46

Earlier quoted context omitted.

Theory is still needed if you want to understand things like variational inference (which is in turn needed to understand things like diffusion models). It’s just like physics - you need math theories to understand things like quantum mechanics, because otherwise it might not make sense.

I think machine learning research is more like engineering, where you do need some math, but you don't need a physics degree. You don't need to understand everything first to discover that some engineering solutions work and others don't. And most abstract theories likely wouldn't have helped you anyway because they are not sufficiently concrete to apply to what you are doing in practice.

To make some progress in ML you might not need a lot of theory, but to understand why things work – you absolutely do. Moreover, the DL field as a whole desperately needs theories explaining what’s going on in these large models.
Post reply on HN