def train_predictor(self, x_old, x_new):
cmin_x = np.min(x_old) - 10 * np.std(x_old)
cmax_x = np.max(x_old) + 10 * np.std(x_old)
x_old = normalize(x_old, cmin_x, cmax_x)
x_new = normalize(x_new, cmin_x, cmax_x)
x_new = preprocess_data(x_new)
x_old = preprocess_data(x_old)
self.Predictor.train_step(np.expand_dims(x_old, 3),
np.expand_dims(x_new, 3),
L=50)
def train_noise_predictor(self, z_new):
cmin_z = np.min(z_new) - 3 * np.std(z_new)
cmax_z = np.max(z_new) + 3 * np.std(z_new)
z_sample = sample(z_new, self.p_len)
z_sample_normalized = normalize(z_sample, cmin_z, cmax_z)
z_new_normalized = normalize(z_new, cmin_z, cmax_z)
z_new_normalized = preprocess_data(z_new_normalized)
z_sample_normalized = preprocess_data(z_sample_normalized)
self.NoisePredictor.train_step(np.expand_dims(z_sample_normalized, 3),
np.expand_dims(z_new_normalized, 3),
L=50,
loss_type="l2")
def train_updator(self, x_old, x_new, z_new):
cmin_z = np.min(z_new) - 3 * np.std(z_new)
cmax_z = np.max(z_new) + 3 * np.std(z_new)
if self.H is None:
cmin_x = np.min(x_old) - 10 * np.std(x_old)
cmax_x = np.max(x_old) + 10 * np.std(x_old)
else:
cmin_x = cmin_z / self.H
cmax_x = cmax_z / self.H
z_new = normalize(z_new, cmin_z, cmax_z)
x_old = normalize(x_old, cmin_x, cmax_x)
x_new = normalize(x_new, cmin_x, cmax_x)
x_new = preprocess_data(x_new)
x_old = preprocess_data(x_old)
x_new_hat = self.Predictor.generator(np.expand_dims(x_old, 3),
training=False)[0].numpy()
x_new_hat = np.reshape(x_new_hat, (-1))
input_data_update = preprocess_Bayesian_data(x_new_hat, z_new)
self.Updator.train_step(np.expand_dims(input_data_update, 0),
np.expand_dims(x_new, 3),
L=50,
loss_type=self.loss_type)
def predict_var(self, x_old, z_new, Ns=30):
cmin_z = np.min(z_new) - 3 * np.std(z_new)
cmax_z = np.max(z_new) + 3 * np.std(z_new)
if self.H is None:
cmin_x = np.min(x_old) - 10 * np.std(x_old)
cmax_x = np.max(x_old) + 10 * np.std(x_old)
else:
cmin_x = cmin_z / self.H
cmax_x = cmax_z / self.H
x_old_normalized = normalize(x_old, cmin_x, cmax_x)
x_old_normalized = preprocess_data(x_old_normalized)
x_new_hat = self.Predictor.generator(np.expand_dims(
x_old_normalized, 3),
training=False)[0].numpy()
x_new_hat = np.reshape(x_new_hat, (-1))
x_stack = []
for ii in range(Ns):
z_sample = sample(z_new, self.p_len)
z_sample_normalized = normalize(z_sample, cmin_z, cmax_z)
z_sample_normalized = preprocess_data(z_sample_normalized)
noise_hat = self.NoisePredictor.generator(
np.expand_dims(z_sample_normalized,
3), training=False)[0].numpy()
noise_hat = np.reshape(noise_hat, (-1))
input_data_update = preprocess_Bayesian_data(
x_new_hat, z_sample_normalized)
x_new_update = self.Updator.generator(np.expand_dims(
input_data_update, 0),
training=False)[0].numpy()
x_new_update = np.reshape(x_new_update, (-1))
x_new_update = unnormalize(x_new_update, cmin_x, cmax_x)
x_stack.append(x_new_update)
x_stack = np.array(x_stack)
var = np.var(x_stack, axis=0)
return var, x_stack
def prop(self, x_old, z_new):
cmin_z = np.min(z_new) - 3 * np.std(z_new)
cmax_z = np.max(z_new) + 3 * np.std(z_new)
if self.H is None:
cmin_x = np.min(x_old) - 10 * np.std(x_old)
cmax_x = np.max(x_old) + 10 * np.std(x_old)
else:
cmin_x = cmin_z / self.H
cmax_x = cmax_z / self.H
x_old = normalize(x_old, cmin_x, cmax_x)
x_old = preprocess_data(x_old)
x_new_hat = self.Predictor.generator(np.expand_dims(x_old, 3),
training=False)[0].numpy()
x_new_hat = np.reshape(x_new_hat, (-1))
x_new_hat = unnormalize(x_new_hat, cmin_x, cmax_x)
return x_new_hat