# incremental gaussian mixture model [closed]

I have trained GMM on small train data set, I would like to update the GMM parameters on the fly when new samples arrive. Please direct on how to do that? Please inform if some existing implementation exits in python

dataset is speech utterance, and I would like to update the parameters of the model of a speaker, as new utterances are added instead of re-training with the entire data.

## closed as too broad by Ben, kjetil b halvorsen, Ferdi, mdewey, Peter Flom♦Dec 27 '18 at 11:35

Please edit the question to limit it to a specific problem with enough detail to identify an adequate answer. Avoid asking multiple distinct questions at once. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

• Can you give more details about the model you are using and the data-set? – mdewey Aug 9 '16 at 8:19
• details are added. Please provide suggestions ! – Shreya Aug 9 '16 at 9:05

Here is an article about an incremental Gaussian Mixture Model (GMM) training: Song and Wang, 2005.

The authors derive formulas for updating the GMM when a new batch of data arrives. Their formulas are interesting in the sense that only newly arrived data is needed to update GMM parameters.

For each new batch, they use covariance and Hotelling's $T^2$ tests to decide if it is needed to merge equivalent clusters.

• thanks for the article! Are you aware of any existing implementation for this ? – Shreya Aug 11 '16 at 11:26
• No, I don't know any. Hope it will help you anyway – Pop Aug 11 '16 at 12:59