Imagine an episode of the Twilight Zone where you wake up tomorrow and there is no such thing as Machine Learning. There are no ML departments, classes, books, experts, etc. Many of the fundamental theoretical components exist but are not unified into a single body of knowledge. What set of subjects would you study to arrive at the same place and depth?
Another angle on the question: Assume that Barrack Obama is fluent in linear algebra, calculus, and probability/statistics. Your goal is to get him to a solid and deep understanding of ML. You are allowed to give him any textbooks you want except for ML textbooks. What kind of books do you give him?
Let's beat this horse to death with another analogy. If we performed PCA on every ML textbook and course out there, what subjects/concepts would the significant principal components correspond to? (Please take this analogy LOOSELY)
What I had in mind to start the list off are:
- Optimization
- Estimation Theory
- Information Theory
- Learning Theory
First to clarify, I am not looking for a "Top 5" list - in fact I'd prefer to see a big list with a lot of depth. Second, I completely expect the subjects to be related, overlapping, etc. Also, an acceptable answer is not "Get a book on data mining."
Thanks for bearing with this idiocy. I do realize it is an ill-posed question but I really appreciate any constructive or insulting feedback.
tl;dr I drank too much coffee and made bad analogies.
[link][11 comments]