How can I add two Gaussian Mixture models?

5 次查看（过去 30 天）
Rizwan Ishaq
Rizwan Ishaq 2016-1-28
回答： Brendan Hamm 2016-1-28
If I have Guassian Mixture model of 5 mixtures, and I get new data, and want to add this data Gaussian mixture model into existing Guassian mixture model, how can I do this?
GMM=fitgmdist(features_vector,16);
now I have new data such as new features_vector features_vectore_new and calculated GMM for this one as well GMM_new=fitgmdist(features_vectore_new,16);
how can I update the GMM with new GMM_new??
h_GMM=GMM+GMM_new??????????
  1 个评论
Adam
Adam 2016-1-28
You can't add two models together. I'm not quite sure what kind of result you are aiming for.
If you have new data you can combine it with the old data and recreate a GMM based on the new combined data set, but if you create independent GMMs out of each set of data then I'm not sure what you are expecting the result to represent?

请先登录，再进行评论。

请先登录，再回答此问题。

回答（1 个）

Brendan Hamm
Brendan Hamm 2016-1-28
The quick answer is you cannot. The fitting function needs to estimate the means vectors and covariance matrices for each of the Gaussians being used. This requires all of the training data to be incorporated simultaneously. So, you essentially need to fit this model all at once:
GMM = fitgmdist([features_vector;features_vectore_new],16);
Or, something similar depending on how you envision the output.

类别

在 Help Center 和 File Exchange 中查找有关 Statistics and Machine Learning Toolbox 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by