def standard_init(self, data):
""" Standard Initial Estimation for *W* and *sigma*.
each *W* raw is set to the average over the data plus WGN of mean zero
and var *sigma*/4. *sigma* is set to the variance of the data around
the computed mean. *pi* is set to 1./H . Returns a dict with the
estimated parameter set with entries "W", "pi" and "sigma".
"""
comm = self.comm
H = self.H
my_y = data['y']
my_N, D = my_y.shape
assert D == self.D
# Calculate averarge W
W_mean = parallel.allmean(my_y, axis=0, comm=comm) # shape: (D, )
# Calculate data variance
sigma_sq = parallel.allmean((my_y-W_mean)**2, axis=0, comm=comm) # shape: (D, )
sigma_init = np.sqrt(sigma_sq).sum() / D # scalar
# Initial W
noise = sigma_init/4.
W_init = W_mean[:,None] + np.random.normal(scale=noise, size=[D, H]) # shape: (H, D)`
#Create and set Model Parameters, W columns have the same average!
model_params = {'W' : W_init }
if 'pies' in self.to_learn:
model_params['pies'] = np.ones(H) * 1./H
return model_params
def standard_init(self, my_data):
""" Standard Initial of the model parameters.
"""
comm = self.comm
temp_params = CAModel.standard_init(self, my_data)
model_params = {}
model_params['W'] = temp_params['W'].copy()
# Initial pi
pi = comm.bcast(np.random.rand(self.H)) * 0.95
pi[pi < 0.05] = 0.05
model_params['pi'] = pi
my_y = my_data['y']
# Calculate averarge W
W_mean = parallel.allmean(my_y, axis=0, comm=comm) # shape: (D, )
# Calculate data variance
sigma_sq_sq = parallel.allmean((my_y - W_mean)**2, axis=0,
comm=comm) # shape: (D, )
# Calculate sigma_sq
if self.sigma_sq_type == 'full':
model_params['sigma_sq'] = np.diag(
np.diag(sigma_sq_sq)) + (0.001 * np.eye(self.D))
elif self.sigma_sq_type == 'diagonal':
model_params['sigma_sq'] = sigma_sq_sq + 0.001
else:
model_params['sigma_sq'] = np.mean(sigma_sq_sq) + 0.001
if 'mu' in self.to_learn:
mu = comm.bcast(np.random.normal(0, 1, [self.H]))
else:
mu = np.zeros(self.H)
model_params['mu'] = mu
if 'psi_sq' in self.to_learn:
psi_sq_diag = comm.bcast(np.random.rand(self.H)) * 2
psi_sq_diag[psi_sq_diag < 0.05] = 0.05
psi_sq = np.diag(psi_sq_diag)
else:
psi_sq = np.eye(self.H)
model_params['psi_sq'] = psi_sq
model_params = comm.bcast(model_params)
return model_params
def standard_init(self, data):
"""Standard onitial estimation for model parameters.
This implementation each "W" raw is set to the average over the data plus white Gaussian noise of mean zero
and standard deviation "sigma"/4. sigma is set to the variance of the data around
the computed mean. "pi" is set to 1./H . Returns a dict with the
estimated parameter set with entries "W", "pi" and "sigma".
Parameters
----------
data : dict
dataset dictionary. Contains a ndarray of size number of samples x number of features under data['y']
Returns
-------
dict
a dictionary containing the model parameters
"""
comm = self.comm
H = self.H
my_y = data['y']
my_N, D = my_y.shape
assert D == self.D
# Calculate averarge W
W_mean = parallel.allmean(my_y, axis=0, comm=comm) # shape: (D, )
# Calculate data variance
# import ipdb;ipdb.set_trace() ######### Break Point ###########
sigma_sq = parallel.allmean((my_y - W_mean)**2, axis=0,
comm=comm) # shape: (D, )
sigma_init = np.sqrt(sigma_sq).sum() / D # scalar
# Initial W
noise = sigma_init / 4.
# noise = sigma_init * 4.
# shape: (H, D)`
# W_init = W_mean[:, None] + np.random.laplace(scale=noise, size=[D, H])
W_init = W_mean[:, None] + np.random.normal(scale=noise, size=[D, H])
sparsity = 1. - (1. / H)
pi_init = np.random.rand(self.K - 1)
pi_init = (1 - sparsity) * pi_init / pi_init.sum()
pi_init = np.insert(pi_init, self._K_0, sparsity)
# Create and set Model Parameters, W columns have the same average!
model_params = {'W': W_init, 'pi': pi_init, 'sigma': sigma_init}
return model_params
def standard_init(self, data):
""" Standard onitial estimation for model parameters.
This implementation
*W* and *sigma*.
each *W* raw is set to the average over the data plus WGN of mean zero
and var *sigma*/4. *sigma* is set to the variance of the data around
the computed mean. *pi* is set to 1./H . Returns a dict with the
estimated parameter set with entries "W", "pi" and "sigma".
"""
comm = self.comm
H = self.H
my_y = data['y']
my_N, D = my_y.shape
assert D == self.D
# Calculate averarge W
W_mean = parallel.allmean(my_y, axis=0, comm=comm) # shape: (D, )
# Calculate data variance
# import ipdb;ipdb.set_trace() ######### Break Point ###########
sigma_sq = parallel.allmean((my_y - W_mean)**2, axis=0, comm=comm) # shape: (D, )
sigma_init = np.sqrt(sigma_sq).sum()/D # scalar
# Initial W
noise = sigma_init / 4.
# noise = sigma_init * 4.
# shape: (H, D)`
# W_init = W_mean[:, None] + np.random.laplace(scale=noise, size=[D, H])
W_init = W_mean[:, None] + np.random.normal(scale=noise, size=[D, H])
sparsity = 1. - (1./H)
pi_init = np.random.rand(self.K - 1)
pi_init = (1 - sparsity) * pi_init / pi_init.sum()
pi_init = np.insert(pi_init, self.K_0, sparsity)
# Create and set Model Parameters, W columns have the same average!
model_params = {
'W': W_init,
'pi': pi_init,
'sigma': sigma_init
}
return model_params
def test_allmean(self):
D = 10
my_a = np.ones(D)
mean = parallel.allmean(my_a)
self.assertAlmostEqual(mean, 1.0)