8000 Modified GMM initialization to only use linear memory and time in sph… by vlad17 · Pull Request #6720 · scikit-learn/scikit-learn · GitHub
[go: up one dir, main page]
Skip to content

Modified GMM initialization to only use linear memory and time in sph… #6720

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.

Sign up for GitHub

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

Closed
wants to merge 1 commit into from
Closed

Modified GMM initialization to only use linear memory and time in sph… #6720

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 18 additions & 4 deletions sklearn/mixture/gmm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -491,7 +491,12 @@ def _fit(self, X, y=None, do_prediction=False):
print('\tWeights have been initialized.')

if 'c' in self.init_params or not hasattr(self, 'covars_'):
cv = np.cov(X.T) + self.min_covar * np.eye(X.shape[1])
# Only compute the full covariance if full or tied
cv = None
if self.covariance_type in ['full', 'tied']:
cv = np.cov(X.T) + self.min_covar * np.eye(X.shape[1])
else:
cv = np.var(X, axis=0) + self.min_covar
if not cv.shape:
cv.shape = (1, 1)
self.covars_ = \
Expand Down Expand Up @@ -759,15 +764,24 @@ def _validate_covars(covars, covariance_type, n_components):

def distribute_covar_matrix_to_match_covariance_type(
tied_cv, covariance_type, n_components):
"""Create all the covariance matrices from a given template."""
"""Create all the covariance matrices from a given template.

If covariance_type is 'spherical' or 'diag', then tied_cv should be
a vector, representing the diagonal of the initial tied covariance matrix.

Otherwise, tied_cv should be the entire covariance matrix."""
if covariance_type == 'spherical':
cv = np.tile(tied_cv.mean() * np.ones(tied_cv.shape[1]),
assert len(tied_cv.shape) == 1
cv = np.tile(tied_cv.mean() * np.ones(len(tied_cv)),
(n_components, 1))
elif covariance_type == 'tied':
assert len(tied_cv.shape) == 2
cv = tied_cv
elif covariance_type == 'diag':
cv = np.tile(np.diag(tied_cv), (n_components, 1))
assert len(tied_cv.shape) == 1
cv = np.tile(tied_cv, (n_components, 1))
elif covariance_type == 'full':
assert len(tied_cv.shape) == 2
cv = np.tile(tied_cv, (n_components, 1, 1))
else:
raise ValueError("covariance_type must be one of " +
Expand Down
0