I've ended up here after asking this on MetaStackOverflow, as the question of where to send "pure Machine Learning" offtopics remains unclear (to me). My initial guess was that the place to put this stuff is CrossValidated (stats), but here is my analysis on the matter:
There are several possibilities for asking ML questions right now (ordered by site traffic):
CrossValidated:

a) It includes machine learning in its main topics:
Cross Validated is a question and answer site for people interested in
statistics, machine learning, data analysis, data mining, and data
visualization.
b) machine-learning
is the 4th most popular tag in the site (3359 tagged questions).
ComputerScience (beta)

a) It mentions "Machine Learning" in its help page, but not in its main page:
Computer Science Stack Exchange is a question and answer site for
students, researchers and practitioners of computer science.
b) machine-learning
is the 22th most popular tag. The site seems to be more about algorithms.
Computational Science is not related to Machine Learning according to its Help page.
Theoretical Computer Science

a) According to its Help Center,
TCS covers a wide variety of topics
including algorithms, data structures, computational complexity,
parallel and distributed computation, probabilistic computation,
quantum computation, automata theory, information theory,
cryptography, program semantics and verification, machine learning,
computational biology, computational economics, computational
geometry, and computational number theory and algebra.
Work in this field is often distinguished by its emphasis on
mathematical technique and rigor.
On the other hand, it also says:
For questions other than research-level questions in TCS, you may want
to consider the following places to ask:
General Artificial Intelligence — Meta Optimize
Statistics and Data Mining — Cross Validated
...
What I understand of all this is (and the word "Theoretical" seems like a hint) is that TCS is the right place for theoretical machine learning questions, leaving practical questions on the side.
b) machine-learning
is the 24th most popular tag in the site.
Data Science

This claims to be a place for "machine learning professionals", and machine-learning
is the most popular tag! (263 questions).
Given all this information, I see there are 2 different options:
a) If the question is practical ("How do I split my 5000 sample training set to improve my SVM performance"), I see two possibilities: CrossValidated and Data Science. Since the final goal is to get help, I'd rather go for CrossValidated as it has a considerable traffic and machine learning seems like a relevant topic in there.
b) If the question is about machine learning theory (for instance, understanding how a neural network or SVM works), there are two natural options: Computer Science or Theoretical Computer Science. If the question is research-level and theoretical (e.g., questions about PAC learning, provable performance guarantees), ask on Theoretical Computer Science. If the question is not research-level, ask on Computer Science.