ylim=flags['ylim'],
delta=flags['delta_lp'],
output_dir=flags['latent_points_dir'],
fig_ext=flags['lp_plotter_fig_ext'])
ls_plotter = LatentSamplePlotter(
image_shape=flags['img_shape'],
xlim=flags['xlim'],
ylim=flags['ylim'],
delta=flags['delta_ls'],
num_samples_to_average=flags['num_samples_to_average'],
output_dir=flags['latent_samples_dir'])
training_data = Data(flags['training_data'],
shuffle_first=flags['shuffle_data'],
batch_size=flags['training_batch_size'],
log_epochs=flags['data_log_epochs'],
name='TrainingData')
Y = training_data.next_batch()
config = tf.ConfigProto(gpu_options=tf.GPUOptions(allow_growth=True))
session = tf.Session(config=config)
kern = gplvm.SEKernel(session=session,
alpha=flags['kernel_alpha'],
gamma=flags['kernel_gamma'],
ARD=flags['kernel_ard'],
Q=flags['q'])
layer = gplvm.GPLVM(Y=Y,
module_spec = importlib.util.spec_from_file_location('flags', 'flags.py')
module = importlib.util.module_from_spec(module_spec)
module_spec.loader.exec_module(module)
flags = module.flags
# Note, this script should be run after the model is trained (run train.py first).
# Basically, the generalisation experiment consist in trying to "reconstruct" the
# test data as closely as possible. The output of this script will be two Numpy arrays,
# one simply containing the test data and the other one the corresponding data generated
# by the model.
init_logging(flags['test_log_file'])
test_data = Data(flags['test_data'],
shuffle_first=False,
batch_size=flags['test_generalisation_batch_size'],
log_epochs=flags['data_log_epochs'],
name='TestData')
test_data_batch = test_data.next_batch()
config = tf.ConfigProto(gpu_options=tf.GPUOptions(allow_growth=True))
session = tf.Session(config=config)
layer1 = SBM_Lower(session=session,
side=flags['img_shape'][0],
side_overlap=flags['layer_1_side_overlap'],
num_h=flags['layer_1_num_h'],
name=flags['layer_1_name'])
layer1.restore(flags['layer_1_ckpt'])