0
$\begingroup$

The Gaussian mixture model is a parametric model that learns the underlying distribution of data. If data contains the noisy attributes and noisy samples then will the model learn that noisy data too or it learns only normal samples and attributes ??? Can we avoid noisy attributes without dimension reduction or feature selection

Is it possible to avoid noisy or redundant attributes while learning the distribution directly??

Here I'm assuming my data to be high dimensional which have redundant and erroneous attributes

$\endgroup$
2
  • $\begingroup$ What kind of noise are you referring too? Gaussian Mixture Model assumes the components to be Gaussian. A Gaussian distribution can be the result of 'noise'. But also many noise will create distributions that are not Gaussian $\endgroup$ – Jon Nordby Jun 25 '20 at 4:05
  • $\begingroup$ In a dataset, if we have redundant/noisy features, should we remove them first and then do the gaussian mixture to learn the distribution, or we can directly use the dataset to learn the distribution? Will noisy features or redundant features change the components of the model ?? Here im assuming my data to be high dimensional which have redundant and erroneous attributes $\endgroup$ – Shivanisrivarshini Jun 25 '20 at 4:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.