def __init__(self,
min_components=3,
max_step_components=30,
max_components=60,
a_split=0.8,
forgetting_factor=0.05,
plot=False,
plot_dims=[0, 1]):
GMM.__init__(self, n_components=min_components, covariance_type='full')
self.params = {
'init_components': min_components,
'max_step_components': max_step_components,
'max_components': max_components,
'a_split': a_split,
'plot': plot,
'plot_dims': plot_dims,
'forgetting_factor': forgetting_factor
}
self.type = 'IGMM'
self.initialized = False
if self.params['plot']:
self.fig_old, self.ax_old = plt.subplots(1, 3)
self.fig_old.suptitle("Incremental Learning of GMM")
self.ax_old[0].set_title('Old Model')
self.ax_old[1].set_title('Short Term Model')
self.ax_old[2].set_title('Current Term Model')
self.fig_old.show()
def __init__(self,
min_components=3,
max_step_components=30,
max_components=60,
a_split=0.8,
forgetting_factor=DynamicParameter(0.05),
x_dims=None,
y_dims=None):
GMM.__init__(self, n_components=min_components, covariance_type='full')
if isinstance(forgetting_factor, float):
forgetting_factor = DynamicParameter(forgetting_factor)
self.params = {
'init_components': min_components,
'max_step_components': max_step_components,
'max_components': max_components,
'a_split': a_split,
'forgetting_factor': forgetting_factor,
'x_dims': x_dims,
'y_dims': y_dims,
'infer_fixed': False
}
if x_dims is not None and y_dims is not None:
self.params['infer_fixed'] = True
self.type = 'IGMM'
self.initialized = False
def __init__(self,
min_components=3,
max_components=10,
distance_method="Kullback-leiber",
merge_type="moment_presaving"):
GaussianMixture.__init__(self,
n_components=min_components,
covariance_type='full')
self.min_components = min_components
self.max_components = max_components
self.type = 'CMGMM'
self.distance_method = distance_method #Kullback-leiber,ISD, Jensen-shannon
self.merge_type = merge_type #moment_presaving,isomorphic
self.initialized = False
self.verbose = False
def __init__(self,
min_components=3,
max_step_components=30,
max_components=60,
a_split=0.8,
forgetting_factor=DynamicParameter(0.05),
x_dims=None,
y_dims=None,
plot=False,
plot_dims=[0, 1]):
GMM.__init__(self, n_components=min_components, covariance_type='full')
if isinstance(forgetting_factor, float):
forgetting_factor = DynamicParameter(forgetting_factor)
self.params = {
'init_components': min_components,
'max_step_components': max_step_components,
'max_components': max_components,
'a_split': a_split,
'plot': plot,
'plot_dims': plot_dims,
'forgetting_factor': forgetting_factor,
'x_dims': x_dims,
'y_dims': y_dims,
'infer_fixed': False
}
if x_dims is not None and y_dims is not None:
self.params['infer_fixed'] = True
self.type = 'IGMM'
self.initialized = False
if self.params['plot']:
self.fig_old, self.ax_old = plt.subplots(1, 3)
self.fig_old.suptitle("Incremental Learning of GMM")
self.ax_old[0].set_title('Old Model')
self.ax_old[1].set_title('Short Term Model')
self.ax_old[2].set_title('Current Term Model')
self.fig_old.show()
def __init__(self, *args, values=None, **kwargs):
""" Instantiate Gaussian mixture model. """
self.values = values
GaussianMixture.__init__(self, *args, **kwargs)
