n_layers_trained = n_DAMP_layers
theta = [None] * n_layers_trained
for iter in range(n_layers_trained):
with tf.variable_scope("Iter" + str(iter)):
theta_thisIter = LDAMP.init_vars_DnCNN(init_mu, init_sigma)
theta[iter] = theta_thisIter
## Construct model
y_measured = LDAMP.GenerateNoisyCSData_handles(x_true, A_handle, sigma_w,
A_val_tf)
if alg == 'DAMP':
(x_hat, MSE_history, NMSE_history, PSNR_history, r, rvar,
dxdr) = LDAMP.LDAMP(y_measured,
A_handle,
At_handle,
A_val_tf,
theta,
x_true,
tie=tie_weights)
elif alg == 'DIT':
(x_hat, MSE_history, NMSE_history,
PSNR_history) = LDAMP.LDIT(y_measured,
A_handle,
At_handle,
A_val_tf,
theta,
x_true,
tie=tie_weights)
else:
raise ValueError('alg was not a supported option')
n_layers_trained = 1
else:
n_layers_trained = n_DAMP_layers
theta = [None] * n_layers_trained
for iter in range(n_layers_trained):
with tf.variable_scope("Iter" + str(iter)):
theta_thisIter = LDAMP.init_vars_DnCNN(init_mu, init_sigma)
theta[iter] = theta_thisIter
## Construct the measurement model and handles/placeholders
[A_handle, At_handle, A_val, A_val_tf] = LDAMP.GenerateMeasurementOperators(measurement_mode)
y_measured = LDAMP.GenerateNoisyCSData_handles(x_true, A_handle, sigma_w, A_val_tf)
## Construct the reconstruction model
if alg=='DAMP':
(x_hat, MSE_history, NMSE_history, PSNR_history, r_final, rvar_final, div_overN) = LDAMP.LDAMP(y_measured,A_handle,At_handle,A_val_tf,theta,x_true,tie=tie_weights,training=training_tf,LayerbyLayer=LayerbyLayer)
elif alg=='DIT':
(x_hat, MSE_history, NMSE_history, PSNR_history) = LDAMP.LDIT(y_measured,A_handle,At_handle,A_val_tf,theta,x_true,tie=tie_weights,training=training_tf,LayerbyLayer=LayerbyLayer)
else:
raise ValueError('alg was not a supported option')
## Define loss and determine which variables to train
nfp = np.float32(height_img * width_img)
if loss_func=='SURE':
assert alg=='DAMP', "Only LDAMP supports training with SURE"
cost = LDAMP.MCSURE_loss(x_hat, div_overN, r_final, tf.sqrt(rvar_final))
elif loss_func=='GSURE':
assert alg == 'DAMP', "Only LDAMP currently supports training with GSURE"
temp0=tf.matmul(A_val_tf,A_val_tf,transpose_b=True)
temp1=tf.matrix_inverse(temp0)
pinv_A=tf.matmul(A_val_tf,temp1,transpose_a=True)