I've got a homework problem in which I have to run Expectation Maximization on 2D Gaussian Mixture Models.
I've gone over half a dozen sets of slides found around the web on the EM algorithm, including those for my class, and they all say pretty much the same thing.....but I'm just not really getting a few fundamental aspects of the equations. Not enough to turn them into code, anyway.
It doesn't help that all the equations seem to be based around 1D Gaussians, and it's left "as an exercise" to figure out how the hell to extend that to 2D.
Anyone know a good "for dummies" reference on this? I can follow the math pretty well *if* it's all laid out----no assumptions on what "should be obvious", etc.
Failing that, anyone understand the alg well enough to answer some specific questions I have?
