Ask HN: What materials to consume to gain knowledge about cutting edge ML?

[smarkelon]

I am a computer science researcher, but do not work in ML. I am interested in gaining a deeper understanding of the cutting edge work being done.

I have taken a graduate level Applied Machine learning course that covered a variety of topics from a cursory theoretical level and then completed programming assignments based on those topics. We covered fundamentals, neural networks, and some selected topics (adversarial ML, privacy attacks, explainable ML, etc.).

If I had a free day to consume materials, what you recommend I read, watch, and play around with to gain a level of understanding past what was covered in this course (with the hope of touching upon the cutting edge of today)?

[efm]

I read a new paper a day with a study group, and when I do not understand something, look at the prior work references. A well written paper is an introduction to the area, and few will help catch you up. Read them closely, there are clues in the details. Then attempt to implement the algorithm, on a tiny example. It's slow going in the beginning but you will catch up. A study group helps.

[kulikalov]

Andrej Karpathy videos on YouTube and Stanford NLP book.