I encountered the term "Global Linear Model" which has the same abbreviation as "Generalized Linear Model".
Since I couldn't find useful information on the former, my question is whether the two concepts are actually the same?
$\begingroup$
$\endgroup$
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
3
Yes, a quick search in google turns up things like this up, and yes, they really do use "GLM" to stand for "global linear model".
There's also the general linear model, which some packages (and books) call "GLM".
None of the three GLMs are the same thing.
Why machine learning people would choose to do that when there's already a degree of confusion over the term "GLM" in statistics ... I can only shake my head.
The title asks for a convention; while "generalized linear model" is undoubtedly the most common (at least for the present), in the end words mean what people use them to mean, and all three appear to be used; we're left with ambiguity that when it crops up must be resolved either from context or by asking for additional information.
-
2$\begingroup$ "Why machine learning people would choose to do that..." Perfectly reasonable if one wants to obfuscate that it is not that new or innovative at all. $\endgroup$– MomoCommented Jul 22, 2015 at 22:56
-
$\begingroup$ @Momo For the uninitiated... What's the underlying concept which is "not new at all"? $\endgroup$– Sycorax ♦Commented Jul 22, 2015 at 23:00
-
1$\begingroup$ @user777 Perceptrons are known since the 1950s en.wikipedia.org/wiki/Perceptron $\endgroup$– MomoCommented Jul 22, 2015 at 23:09