I like your suggestion Grant.
One reason for keeping the discussion on the mahout mailing list is
because once you get past theory and into implementation, I think a
lot of the discussion will deal with numerical issues that would be of
interest to mahout users/developers.
Just my two sense.
J
Quoting Grant Ingersoll <[email protected]>:
On a related note, I'd love to see/encourage more here on this list
to discuss foundational aspects of machine learning (ML), such as
Lin. Algebra, Stats, AI, etc. even if they aren't specifically
Mahout related. Perhaps people could prefix the subject line with
the topic, as in: [Linear Algebra] EigenValue Decomposition
questions or [Stats] Bayes Theorem Question. Personally, I find the
math, etc. to be fairly easy to follow when presented in the
abstract, but sometimes struggle with what it means in the context
of ML.
Alternatively, I know some here are likely watching the Stanford CS
229 lectures, so those could be prefixed by [CS229-lec 5] or the
Strang Lin. Algebra lectures or any of the other things on our
https://cwiki.apache.org/confluence/display/MAHOUT/Reference+Reading
list
Just a suggestion, what do others think? Does it encourage too much
noise? Not useful?
(We could create a new mailing list for these things, but I'd
suggest we just start here first and if it gets too noisy, we can
move it.)
-Grant
On Sep 5, 2010, at 11:32 PM, Lance Norskog wrote:
The Hadoop lists seem to be all about the sysad aspects of Hadoop,
while Mahout users talk about algorithms a lot more. Are there
other hangouts oriented towards M/R or other large-scale algorithm
discussion?
Thanks,
Lance
p.s. by 'sysad' I mean 'Google Envy'.
