cleaned up density structure, use parent mixture class to organize model variations

Author: Alexander Ororbia

docs/source/ngclearn.utils.density.rst

@@ -4,18 +4,18 @@ ngclearn.utils.density package
 Submodules
 ----------
 
-ngclearn.utils.density.bmm module
----------------------------------
+ngclearn.utils.density.bernoulliMixture module
+----------------------------------------------
 
-.. automodule:: ngclearn.utils.density.bmm
+.. automodule:: ngclearn.utils.density.bernoulliMixture
    :members:
    :undoc-members:
    :show-inheritance:
 
-ngclearn.utils.density.gmm module
----------------------------------
+ngclearn.utils.density.gaussianMixture module
+---------------------------------------------
 
-.. automodule:: ngclearn.utils.density.gmm
+.. automodule:: ngclearn.utils.density.gaussianMixture
    :members:
    :undoc-members:
    :show-inheritance:

docs/tutorials/neurocog/density_modeling.md

@@ -11,7 +11,7 @@ The following Python code will employ an NGC-Learn-in-built GMM density estimato
 
 ```python
 from jax import numpy as jnp, random
-from ngclearn.utils.density.gmm import GMM ## pull out the mixture model density estimator
+from ngclearn.utils.density.gaussianMixture import GaussianMixture as GMM ## pull out density estimator
 
 def gen_data(dkey, n_samp_per_mode): ## data generator (or proxy stochastic data generating process)
     scale = 0.3

ngclearn/utils/density/__init__.py

@@ -1,5 +1,5 @@
 from .mixture import Mixture ## general mixture template parent class
 ## point to supported density estimator models
-from .gmm import GMM ## Gaussian mixture model 
-from .bmm import BMM ## Bernoulli mixture model
+from .gaussianMixture import GaussianMixture ## Gaussian mixture model 
+from .bernoulliMixture import BernoulliMixture ## Bernoulli mixture model
 

ngclearn/utils/density/bmm.py …clearn/utils/density/bernoulliMixture.pyngclearn/utils/density/bmm.py renamed to ngclearn/utils/density/bernoulliMixture.py ngclearn/utils/density/bmm.py renamed to ngclearn/utils/density/bernoulliMixture.py

@@ -60,7 +60,7 @@ def _sample_component(dkey, n_samples, mu): ## samples a component (of mixture)
 
 ########################################################################################################################
 
-class BMM(Mixture): ## Bernoulli mixture model (mixture-of-Bernoullis)
+class BernoulliMixture(Mixture): ## Bernoulli mixture model (mixture-of-Bernoullis)
     """
     Implements a Bernoulli mixture model (BMM) -- or mixture of Bernoullis (MoB).
     Adaptation of parameters is conducted via the Expectation-Maximization (EM)

ngclearn/utils/density/gmm.py ngclearn/utils/density/gaussianMixture.pyngclearn/utils/density/gmm.py renamed to ngclearn/utils/density/gaussianMixture.py

@@ -101,7 +101,7 @@ def _sample_component(dkey, n_samples, mu, Sigma, assume_diag_cov=False): ## sam
 
 ########################################################################################################################
 
-class GMM(Mixture): ## Gaussian mixture model (mixture-of-Gaussians)
+class GaussianMixture(Mixture): ## Gaussian mixture model (mixture-of-Gaussians)
     """
     Implements a Gaussian mixture model (GMM) -- or mixture of Gaussians (MoG).
     Adaptation of parameters is conducted via the Expectation-Maximization (EM)