Re: EM ALGORITHM

Hi,

I am guessing you are looking for the Expectation Maximization (EM) Algorithm ? A google search will bring up a lot of good links. Also wikipedia has a section on this
http://en.wikipedia.org/wiki/Expectation-maximization_algorithm

Best
Ajay


mlimber wrote: 
sabbathian@xxxxxxxxx wrote:

mlimber wrote:

sabbathian@xxxxxxxxx wrote:

DOES ANYONE HAVE OR KNOW WHERE I CAN FIND A GOOD AND EASY
TUTORIAL FOR
THE EM ALGORITHM WITH EXAMPLES??

THANKS! 

Which EM algorithm are you refering to, and why are you yelling at us?

Cheers! --M 

I am sorry about the caps.

**Please note, that this is not a homework question.
I am looking for an EM algorithm that can be used for missing
features/data/values?

E.g.
The following is a hypothetical example:

assume x to be a missing feature/data/value


                 f1        f2       f3         f4           f5
f6        f7       f8
first point = [0.33    0.045  0.454   1.556     x        0.8      x
 1.2]
sec point = [0.23    x        0.980    0.64      2.34   x         0.7
 1.0]
third point= [x         0.2     1.200      0.98     8.2    1.2
1.67   1.3]
.
.
.
n-th point=[0.35      0.023  0.444   0.78      x         1.11  0.89
0.9]

hence my total dataset is n by 8. The 8 columns each represent a
reading of perhaps a sensor. Each sensor may be reading a different
feature value. (e.g. gas value, chemical index)


How can the EM be used? Furthermore, the problem becomes more
complicated if each of the datapoints can be classified as one of m
classes( class1 , class2 ,.... ,classm). So we do not know the
distribution (i.e. we do not know which class each of the points belong
to)

A good tutorial or explanation on how to approach the problem will be
helpful. Also note, that there can be any number of features and points
and are not restricted to 8 features as shown above.



I still don't know what "EM" stands for in this context (element
matching? ergodic morphology? end mean? elephant march?). Also, I
suspect your question has less to do with image processing proper and
more with pattern matching and classification. If so, you'd probably be
better served in another newsgroup. (If you were concerned about the
extraction of features rather than their classification, you'd be in
the right place, however.)

Cheers! --M

.

Relevant Pages

  • Re: EM ALGORITHM
    ... I am looking for an EM algorithm that can be used for missing ... Each sensor may be reading a different ... that there can be any number of features and points ...
    (sci.image.processing)
  • Re: EM ALGORITHM
    ... The EM algorithm is known as Expectation Maximization. ... that there can be any number of features and points ... > more with pattern matching and classification. ...
    (sci.image.processing)
  • Re: EM ALGORITHM
    ... The EM algorithm is known as Expectation Maximization. ... that there can be any number of features and points ... > more with pattern matching and classification. ...
    (sci.image.processing)
  • Re: EM ALGORITHM
    ... >> Which EM algorithm are you refering to, and why are you yelling at us? ... that there can be any number of features and points ... more with pattern matching and classification. ...
    (sci.image.processing)
  • Re: Help on choosing a valid pattern: composite or not?
    ... when you have a set of features x1,x2,..,xN, a set of classes ... Once weights in all nodes have stabilized, ... >> I moved that into the training and classification contexts. ...
    (comp.object)