Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

start work on gmm #18

Draft
wants to merge 2 commits into
base: main
Choose a base branch
from
Draft

start work on gmm #18

wants to merge 2 commits into from

Conversation

ezhang94
Copy link

@ezhang94 ezhang94 commented Dec 10, 2021

Adding in GaussianMixture distribution with E-M fitting functionality, tests, and notebook demonstrations.

Maximum likelihood estimation of MVN distribution

  • Unweighted ML estimation of MVN full-covariance distribution
  • Weighted ML estimation of MVN full-covariance distribution
  • Unweighted and weighted ML estimation of MVN diagonal cov distribution
  • Unweighted and weighted ML estimation of MVN spherical cov distribution

Expectation maximization inference

  • Create barebones GaussianMixture distribution and sample
  • Test EM inference with full-covariance MVN distribution
  • + diagonal-covariance MVN components distribution
  • + spherical-covariance MVN components distribution

Sidequests

  • Compare home-brewed k-means implementation (see gimbal-refactor repo) with sklearn.cluster.KMeans (additional benefits: does not require reshaping dataset. perhaps can generalize to different distances, i.e. cosine distance to permit generalization to spherical k-means)
  • Propose simpler one_hot implementation

@schlagercollin
Copy link
Member

Nice!

@ezhang94 ezhang94 self-assigned this Dec 10, 2021
Verified and tested for MultivariateNormalFullCovariance distribution
@lindermanlab lindermanlab deleted a comment from github-actions bot Dec 14, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants