2

I am working on biomedical signal analysis and the most used method for parameters estimation is the EM algorithm.

My question is : what are the most powerful alternatives to this algorithm?

WildThing
  • 141
  • 1
  • 1
    What exactly are you using EM for? This algorithm is has really many uses, so you have to be more precise. – Tim May 15 '15 at 12:57
  • @Tim I am using it for estimating the parameters of different normally distributed classes. – WildThing May 15 '15 at 13:16
  • @Tim It is a gaussian mixture model and the data is brain MRI scans. – WildThing May 15 '15 at 13:20
  • 1
    OK, now it is more clear. As far as I know EM is most commonly used in this area, but Bayesian estimation is also possible - have you considered it? – Tim May 15 '15 at 13:23
  • @Tim yes, i know that it is possible to use EM with the MAP criterion (which is a kind of Bayesian reasoning). But i am looking for an alternative. – WildThing May 15 '15 at 13:29
  • 1
    Maybe you should edit your question to provide additional details. Why EM is not good for you? As far as I know EM is a state-of-art in this area, so maybe say why it is not satisfactory? – Tim May 15 '15 at 13:36
  • @Tim I am not saying that it is not satisfactory, i am just looking for something different to try. – WildThing May 15 '15 at 13:41
  • 2
    Then you could probably try Bayesian MCMC, but others may provide a more detailed answer. BTW, consider editing your question to add all those deatils that came out in our discussion so others know better what are you looking for. – Tim May 15 '15 at 13:42

0 Answers0