def _run(model_file_name, target_class, target_layer_name,
top_example_dir_name, storm_metafile_name, num_examples,
randomize_weights, cascading_random, output_file_name):
"""Runs Grad-CAM (gradient-weighted class-activation maps).
This is effectively the main method.
:param model_file_name: See documentation at top of file.
:param target_class: Same.
:param target_layer_name: Same.
:param top_example_dir_name: Same.
:param storm_metafile_name: Same.
:param num_examples: Same.
:param randomize_weights: Same.
:param cascading_random: Same.
:param output_file_name: Same.
"""
file_system_utils.mkdir_recursive_if_necessary(file_name=output_file_name)
# Read model and metadata.
print('Reading model from: "{0:s}"...'.format(model_file_name))
model_object = cnn.read_model(model_file_name)
model_metafile_name = '{0:s}/model_metadata.p'.format(
os.path.split(model_file_name)[0])
print(
'Reading model metadata from: "{0:s}"...'.format(model_metafile_name))
model_metadata_dict = cnn.read_model_metadata(model_metafile_name)
training_option_dict = model_metadata_dict[cnn.TRAINING_OPTION_DICT_KEY]
training_option_dict[trainval_io.REFLECTIVITY_MASK_KEY] = None
output_dir_name, pathless_output_file_name = os.path.split(
output_file_name)
extensionless_output_file_name, output_file_extension = os.path.splitext(
pathless_output_file_name)
if randomize_weights:
conv_dense_layer_names = _find_conv_and_dense_layers(model_object)
conv_dense_layer_names.reverse()
num_sets = len(conv_dense_layer_names)
else:
conv_dense_layer_names = []
num_sets = 1
print(
'Reading storm metadata from: "{0:s}"...'.format(storm_metafile_name))
full_storm_id_strings, storm_times_unix_sec = (
tracking_io.read_ids_and_times(storm_metafile_name))
print(SEPARATOR_STRING)
if 0 < num_examples < len(full_storm_id_strings):
full_storm_id_strings = full_storm_id_strings[:num_examples]
storm_times_unix_sec = storm_times_unix_sec[:num_examples]
example_dict = testing_io.read_predictors_specific_examples(
top_example_dir_name=top_example_dir_name,
desired_full_id_strings=full_storm_id_strings,
desired_times_unix_sec=storm_times_unix_sec,
option_dict=training_option_dict,
layer_operation_dicts=model_metadata_dict[cnn.LAYER_OPERATIONS_KEY])
print(SEPARATOR_STRING)
predictor_matrices = example_dict[testing_io.INPUT_MATRICES_KEY]
sounding_pressure_matrix_pa = (
example_dict[testing_io.SOUNDING_PRESSURES_KEY])
print('Denormalizing model inputs...')
denorm_predictor_matrices = trainval_io.separate_shear_and_reflectivity(
list_of_input_matrices=copy.deepcopy(predictor_matrices),
training_option_dict=training_option_dict)
denorm_predictor_matrices = model_interpretation.denormalize_data(
list_of_input_matrices=denorm_predictor_matrices,
model_metadata_dict=model_metadata_dict)
print(SEPARATOR_STRING)
for k in range(num_sets):
if randomize_weights:
if cascading_random:
_reset_weights_in_layer(model_object=model_object,
layer_name=conv_dense_layer_names[k])
this_model_object = model_object
this_output_file_name = (
'{0:s}/{1:s}_cascading-random_{2:s}{3:s}').format(
output_dir_name, extensionless_output_file_name,
conv_dense_layer_names[k].replace('_', '-'),
output_file_extension)
else:
this_model_object = keras.models.Model.from_config(
model_object.get_config())
this_model_object.set_weights(model_object.get_weights())
_reset_weights_in_layer(model_object=this_model_object,
layer_name=conv_dense_layer_names[k])
this_output_file_name = '{0:s}/{1:s}_random_{2:s}{3:s}'.format(
output_dir_name, extensionless_output_file_name,
conv_dense_layer_names[k].replace('_', '-'),
output_file_extension)
else:
this_model_object = model_object
this_output_file_name = output_file_name
# print(K.eval(this_model_object.get_layer(name='dense_53').weights[0]))
these_cam_matrices, these_guided_cam_matrices = (
_run_gradcam_one_weight_set(
model_object=this_model_object,
target_class=target_class,
target_layer_name=target_layer_name,
predictor_matrices=predictor_matrices,
training_option_dict=training_option_dict))
print('Writing results to file: "{0:s}"...'.format(
this_output_file_name))
gradcam.write_standard_file(
pickle_file_name=this_output_file_name,
denorm_predictor_matrices=denorm_predictor_matrices,
cam_matrices=these_cam_matrices,
guided_cam_matrices=these_guided_cam_matrices,
full_storm_id_strings=full_storm_id_strings,
storm_times_unix_sec=storm_times_unix_sec,
model_file_name=model_file_name,
target_class=target_class,
target_layer_name=target_layer_name,
sounding_pressure_matrix_pa=sounding_pressure_matrix_pa)
print(SEPARATOR_STRING)
def _run(model_file_name, top_example_dir_name, storm_metafile_name,
num_examples, output_file_name):
"""Creates dummy saliency map for each storm object.
This is effectively the main method.
:param model_file_name: See documentation at top of file.
:param top_example_dir_name: Same.
:param storm_metafile_name: Same.
:param num_examples: Same.
:param output_file_name: Same.
"""
file_system_utils.mkdir_recursive_if_necessary(file_name=output_file_name)
model_metafile_name = '{0:s}/model_metadata.p'.format(
os.path.split(model_file_name)[0])
print(
'Reading model metadata from: "{0:s}"...'.format(model_metafile_name))
model_metadata_dict = cnn.read_model_metadata(model_metafile_name)
training_option_dict = model_metadata_dict[cnn.TRAINING_OPTION_DICT_KEY]
training_option_dict[trainval_io.REFLECTIVITY_MASK_KEY] = None
print(
'Reading storm metadata from: "{0:s}"...'.format(storm_metafile_name))
full_storm_id_strings, storm_times_unix_sec = (
tracking_io.read_ids_and_times(storm_metafile_name))
print(SEPARATOR_STRING)
if 0 < num_examples < len(full_storm_id_strings):
full_storm_id_strings = full_storm_id_strings[:num_examples]
storm_times_unix_sec = storm_times_unix_sec[:num_examples]
example_dict = testing_io.read_predictors_specific_examples(
top_example_dir_name=top_example_dir_name,
desired_full_id_strings=full_storm_id_strings,
desired_times_unix_sec=storm_times_unix_sec,
option_dict=training_option_dict,
layer_operation_dicts=model_metadata_dict[cnn.LAYER_OPERATIONS_KEY])
print(SEPARATOR_STRING)
predictor_matrices = example_dict[testing_io.INPUT_MATRICES_KEY]
sounding_pressure_matrix_pa = (
example_dict[testing_io.SOUNDING_PRESSURES_KEY])
radar_matrix = predictor_matrices[0]
num_examples = radar_matrix.shape[0]
num_channels = radar_matrix.shape[-1]
num_spatial_dim = len(radar_matrix.shape) - 2
if num_spatial_dim == 2:
kernel_matrix = numpy.expand_dims(EDGE_DETECTOR_MATRIX_2D, axis=-1)
else:
kernel_matrix = numpy.expand_dims(EDGE_DETECTOR_MATRIX_3D, axis=-1)
kernel_matrix = numpy.repeat(kernel_matrix, num_channels, axis=-1)
kernel_matrix = numpy.expand_dims(kernel_matrix, axis=-1)
kernel_matrix = numpy.repeat(kernel_matrix, num_channels, axis=-1)
radar_saliency_matrix = numpy.full(radar_matrix.shape, numpy.nan)
for i in range(num_examples):
if numpy.mod(i, 10) == 0:
print((
'Have created dummy saliency map for {0:d} of {1:d} examples...'
).format(i, num_examples))
if num_spatial_dim == 2:
this_saliency_matrix = standalone_utils.do_2d_convolution(
feature_matrix=radar_matrix[i, ...],
kernel_matrix=kernel_matrix,
pad_edges=True,
stride_length_px=1)
else:
this_saliency_matrix = standalone_utils.do_3d_convolution(
feature_matrix=radar_matrix[i, ...],
kernel_matrix=kernel_matrix,
pad_edges=True,
stride_length_px=1)
radar_saliency_matrix[i, ...] = this_saliency_matrix[0, ...]
print('Have created dummy saliency map for all {0:d} examples!'.format(
num_examples))
print(SEPARATOR_STRING)
saliency_matrices = [
radar_saliency_matrix if k == 0 else predictor_matrices[k]
for k in range(len(predictor_matrices))
]
saliency_matrices = trainval_io.separate_shear_and_reflectivity(
list_of_input_matrices=saliency_matrices,
training_option_dict=training_option_dict)
denorm_predictor_matrices = trainval_io.separate_shear_and_reflectivity(
list_of_input_matrices=copy.deepcopy(predictor_matrices),
training_option_dict=training_option_dict)
print('Denormalizing model inputs...')
denorm_predictor_matrices = model_interpretation.denormalize_data(
list_of_input_matrices=denorm_predictor_matrices,
model_metadata_dict=model_metadata_dict)
print('Writing saliency maps to file: "{0:s}"...'.format(output_file_name))
saliency_metadata_dict = saliency_maps.check_metadata(
component_type_string=model_interpretation.CLASS_COMPONENT_TYPE_STRING,
target_class=1)
saliency_maps.write_standard_file(
pickle_file_name=output_file_name,
denorm_predictor_matrices=denorm_predictor_matrices,
saliency_matrices=saliency_matrices,
full_storm_id_strings=full_storm_id_strings,
storm_times_unix_sec=storm_times_unix_sec,
model_file_name=model_file_name,
metadata_dict=saliency_metadata_dict,
sounding_pressure_matrix_pa=sounding_pressure_matrix_pa)
def _run_gradcam_one_weight_set(model_object, target_class, target_layer_name,
predictor_matrices, training_option_dict):
"""Runs Grad-CAM with one set of weights.
T = number of input tensors to model
:param model_object: Trained CNN (instance of `keras.models.Model` or
`keras.models.Sequential`).
:param target_class: See documentation at top of file.
:param target_layer_name: Same.
:param predictor_matrices: length-T list of numpy arrays, containing
normalized predictor matrices.
:param training_option_dict: Dictionary returned by
`cnn.read_model_metadata`.
:return: cam_matrices: length-T list of numpy arrays, containing unguided
class activations.
:return: guided_cam_matrices: length-T list of numpy arrays, containing
guided class activations.
"""
num_matrices = len(predictor_matrices)
num_examples = predictor_matrices[0].shape[0]
cam_matrices = [None] * num_matrices
guided_cam_matrices = [None] * num_matrices
new_model_object = None
for i in range(num_examples):
print('Running Grad-CAM for example {0:d} of {1:d}...'.format(
i + 1, num_examples))
these_predictor_matrices = [a[[i], ...] for a in predictor_matrices]
these_cam_matrices = gradcam.run_gradcam(
model_object=model_object,
list_of_input_matrices=these_predictor_matrices,
target_class=target_class,
target_layer_name=target_layer_name)
print('Running guided Grad-CAM for example {0:d} of {1:d}...'.format(
i + 1, num_examples))
these_guided_cam_matrices, new_model_object = (
gradcam.run_guided_gradcam(
orig_model_object=model_object,
list_of_input_matrices=these_predictor_matrices,
target_layer_name=target_layer_name,
list_of_cam_matrices=these_cam_matrices,
new_model_object=new_model_object))
if all([a is None for a in cam_matrices]):
for k in range(num_matrices):
if these_cam_matrices[k] is None:
continue
these_dim = numpy.array(
(num_examples, ) + these_cam_matrices[k].shape[1:],
dtype=int)
cam_matrices[k] = numpy.full(these_dim, numpy.nan)
these_dim = numpy.array(
(num_examples, ) + these_guided_cam_matrices[k].shape[1:],
dtype=int)
guided_cam_matrices[k] = numpy.full(these_dim, numpy.nan)
for k in range(num_matrices):
if these_cam_matrices[k] is None:
continue
cam_matrices[k][i, ...] = these_cam_matrices[k][0, ...]
guided_cam_matrices[k][i,
...] = (these_guided_cam_matrices[k][0,
...])
upsample_refl = training_option_dict[trainval_io.UPSAMPLE_REFLECTIVITY_KEY]
if upsample_refl:
cam_matrices[0] = numpy.expand_dims(cam_matrices[0], axis=-1)
num_channels = predictor_matrices[0].shape[-1]
cam_matrices[0] = numpy.repeat(a=cam_matrices[0],
repeats=num_channels,
axis=-1)
cam_matrices = trainval_io.separate_shear_and_reflectivity(
list_of_input_matrices=cam_matrices,
training_option_dict=training_option_dict)
cam_matrices[0] = cam_matrices[0][..., 0]
cam_matrices[1] = cam_matrices[1][..., 0]
guided_cam_matrices = trainval_io.separate_shear_and_reflectivity(
list_of_input_matrices=guided_cam_matrices,
training_option_dict=training_option_dict)
return cam_matrices, guided_cam_matrices
def _run(model_file_name, component_type_string, target_class, layer_name,
ideal_activation, neuron_indices, channel_index, top_example_dir_name,
storm_metafile_name, num_examples, randomize_weights,
cascading_random, output_file_name):
"""Computes saliency map for each storm object and each model component.
This is effectively the main method.
:param model_file_name: See documentation at top of file.
:param component_type_string: Same.
:param target_class: Same.
:param layer_name: Same.
:param ideal_activation: Same.
:param neuron_indices: Same.
:param channel_index: Same.
:param top_example_dir_name: Same.
:param storm_metafile_name: Same.
:param num_examples: Same.
:param randomize_weights: Same.
:param cascading_random: Same.
:param output_file_name: Same.
"""
# Check input args.
file_system_utils.mkdir_recursive_if_necessary(file_name=output_file_name)
model_interpretation.check_component_type(component_type_string)
# Read model and metadata.
print('Reading model from: "{0:s}"...'.format(model_file_name))
model_object = cnn.read_model(model_file_name)
model_metafile_name = '{0:s}/model_metadata.p'.format(
os.path.split(model_file_name)[0])
print(
'Reading model metadata from: "{0:s}"...'.format(model_metafile_name))
model_metadata_dict = cnn.read_model_metadata(model_metafile_name)
training_option_dict = model_metadata_dict[cnn.TRAINING_OPTION_DICT_KEY]
training_option_dict[trainval_io.REFLECTIVITY_MASK_KEY] = None
output_dir_name, pathless_output_file_name = os.path.split(
output_file_name)
extensionless_output_file_name, output_file_extension = os.path.splitext(
pathless_output_file_name)
if randomize_weights:
conv_dense_layer_names = _find_conv_and_dense_layers(model_object)
conv_dense_layer_names.reverse()
num_sets = len(conv_dense_layer_names)
else:
conv_dense_layer_names = []
num_sets = 1
print(
'Reading storm metadata from: "{0:s}"...'.format(storm_metafile_name))
full_storm_id_strings, storm_times_unix_sec = (
tracking_io.read_ids_and_times(storm_metafile_name))
print(SEPARATOR_STRING)
if 0 < num_examples < len(full_storm_id_strings):
full_storm_id_strings = full_storm_id_strings[:num_examples]
storm_times_unix_sec = storm_times_unix_sec[:num_examples]
example_dict = testing_io.read_predictors_specific_examples(
top_example_dir_name=top_example_dir_name,
desired_full_id_strings=full_storm_id_strings,
desired_times_unix_sec=storm_times_unix_sec,
option_dict=training_option_dict,
layer_operation_dicts=model_metadata_dict[cnn.LAYER_OPERATIONS_KEY])
print(SEPARATOR_STRING)
predictor_matrices = example_dict[testing_io.INPUT_MATRICES_KEY]
sounding_pressure_matrix_pa = example_dict[
testing_io.SOUNDING_PRESSURES_KEY]
denorm_predictor_matrices = trainval_io.separate_shear_and_reflectivity(
list_of_input_matrices=copy.deepcopy(predictor_matrices),
training_option_dict=training_option_dict)
print('Denormalizing model inputs...')
denorm_predictor_matrices = model_interpretation.denormalize_data(
list_of_input_matrices=denorm_predictor_matrices,
model_metadata_dict=model_metadata_dict)
print(SEPARATOR_STRING)
for k in range(num_sets):
if randomize_weights:
if cascading_random:
_reset_weights_in_layer(model_object=model_object,
layer_name=conv_dense_layer_names[k])
this_model_object = model_object
this_output_file_name = (
'{0:s}/{1:s}_cascading-random_{2:s}{3:s}').format(
output_dir_name, extensionless_output_file_name,
conv_dense_layer_names[k].replace('_', '-'),
output_file_extension)
else:
this_model_object = keras.models.Model.from_config(
model_object.get_config())
this_model_object.set_weights(model_object.get_weights())
_reset_weights_in_layer(model_object=this_model_object,
layer_name=conv_dense_layer_names[k])
this_output_file_name = '{0:s}/{1:s}_random_{2:s}{3:s}'.format(
output_dir_name, extensionless_output_file_name,
conv_dense_layer_names[k].replace('_', '-'),
output_file_extension)
else:
this_model_object = model_object
this_output_file_name = output_file_name
# print(K.eval(this_model_object.get_layer(name='dense_3').weights[0]))
if component_type_string == CLASS_COMPONENT_TYPE_STRING:
print('Computing saliency maps for target class {0:d}...'.format(
target_class))
saliency_matrices = (
saliency_maps.get_saliency_maps_for_class_activation(
model_object=this_model_object,
target_class=target_class,
list_of_input_matrices=predictor_matrices))
elif component_type_string == NEURON_COMPONENT_TYPE_STRING:
print(
('Computing saliency maps for neuron {0:s} in layer "{1:s}"...'
).format(str(neuron_indices), layer_name))
saliency_matrices = (
saliency_maps.get_saliency_maps_for_neuron_activation(
model_object=this_model_object,
layer_name=layer_name,
neuron_indices=neuron_indices,
list_of_input_matrices=predictor_matrices,
ideal_activation=ideal_activation))
else:
print((
'Computing saliency maps for channel {0:d} in layer "{1:s}"...'
).format(channel_index, layer_name))
saliency_matrices = (
saliency_maps.get_saliency_maps_for_channel_activation(
model_object=this_model_object,
layer_name=layer_name,
channel_index=channel_index,
list_of_input_matrices=predictor_matrices,
stat_function_for_neuron_activations=K.max,
ideal_activation=ideal_activation))
saliency_matrices = trainval_io.separate_shear_and_reflectivity(
list_of_input_matrices=saliency_matrices,
training_option_dict=training_option_dict)
print('Writing saliency maps to file: "{0:s}"...'.format(
this_output_file_name))
saliency_metadata_dict = saliency_maps.check_metadata(
component_type_string=component_type_string,
target_class=target_class,
layer_name=layer_name,
ideal_activation=ideal_activation,
neuron_indices=neuron_indices,
channel_index=channel_index)
saliency_maps.write_standard_file(
pickle_file_name=this_output_file_name,
denorm_predictor_matrices=denorm_predictor_matrices,
saliency_matrices=saliency_matrices,
full_storm_id_strings=full_storm_id_strings,
storm_times_unix_sec=storm_times_unix_sec,
model_file_name=model_file_name,
metadata_dict=saliency_metadata_dict,
sounding_pressure_matrix_pa=sounding_pressure_matrix_pa)
def _run(model_file_name, target_class, target_layer_name,
top_example_dir_name, storm_metafile_name, num_examples,
output_file_name):
"""Runs Grad-CAM (gradient-weighted class-activation maps).
This is effectively the main method.
:param model_file_name: See documentation at top of file.
:param target_class: Same.
:param target_layer_name: Same.
:param top_example_dir_name: Same.
:param storm_metafile_name: Same.
:param num_examples: Same.
:param output_file_name: Same.
"""
file_system_utils.mkdir_recursive_if_necessary(file_name=output_file_name)
# Read model and metadata.
print('Reading model from: "{0:s}"...'.format(model_file_name))
model_object = cnn.read_model(model_file_name)
model_metafile_name = '{0:s}/model_metadata.p'.format(
os.path.split(model_file_name)[0])
print(
'Reading model metadata from: "{0:s}"...'.format(model_metafile_name))
model_metadata_dict = cnn.read_model_metadata(model_metafile_name)
training_option_dict = model_metadata_dict[cnn.TRAINING_OPTION_DICT_KEY]
training_option_dict[trainval_io.REFLECTIVITY_MASK_KEY] = None
print(
'Reading storm metadata from: "{0:s}"...'.format(storm_metafile_name))
full_id_strings, storm_times_unix_sec = tracking_io.read_ids_and_times(
storm_metafile_name)
print(SEPARATOR_STRING)
if 0 < num_examples < len(full_id_strings):
full_id_strings = full_id_strings[:num_examples]
storm_times_unix_sec = storm_times_unix_sec[:num_examples]
list_of_input_matrices, sounding_pressure_matrix_pascals = (
testing_io.read_specific_examples(
top_example_dir_name=top_example_dir_name,
desired_full_id_strings=full_id_strings,
desired_times_unix_sec=storm_times_unix_sec,
option_dict=training_option_dict,
list_of_layer_operation_dicts=model_metadata_dict[
cnn.LAYER_OPERATIONS_KEY]))
print(SEPARATOR_STRING)
list_of_cam_matrices = None
list_of_guided_cam_matrices = None
new_model_object = None
num_examples = len(full_id_strings)
for i in range(num_examples):
print('Running Grad-CAM for example {0:d} of {1:d}...'.format(
i + 1, num_examples))
these_input_matrices = [a[[i], ...] for a in list_of_input_matrices]
these_cam_matrices = gradcam.run_gradcam(
model_object=model_object,
list_of_input_matrices=these_input_matrices,
target_class=target_class,
target_layer_name=target_layer_name)
print('Running guided Grad-CAM for example {0:d} of {1:d}...'.format(
i + 1, num_examples))
these_guided_cam_matrices, new_model_object = (
gradcam.run_guided_gradcam(
orig_model_object=model_object,
list_of_input_matrices=these_input_matrices,
target_layer_name=target_layer_name,
list_of_cam_matrices=these_cam_matrices,
new_model_object=new_model_object))
if list_of_cam_matrices is None:
list_of_cam_matrices = copy.deepcopy(these_cam_matrices)
list_of_guided_cam_matrices = copy.deepcopy(
these_guided_cam_matrices)
else:
for j in range(len(these_cam_matrices)):
if list_of_cam_matrices[j] is None:
continue
list_of_cam_matrices[j] = numpy.concatenate(
(list_of_cam_matrices[j], these_cam_matrices[j]), axis=0)
list_of_guided_cam_matrices[j] = numpy.concatenate(
(list_of_guided_cam_matrices[j],
these_guided_cam_matrices[j]),
axis=0)
print(SEPARATOR_STRING)
upsample_refl = training_option_dict[trainval_io.UPSAMPLE_REFLECTIVITY_KEY]
if upsample_refl:
list_of_cam_matrices[0] = numpy.expand_dims(list_of_cam_matrices[0],
axis=-1)
num_channels = list_of_input_matrices[0].shape[-1]
list_of_cam_matrices[0] = numpy.repeat(a=list_of_cam_matrices[0],
repeats=num_channels,
axis=-1)
list_of_cam_matrices = trainval_io.separate_shear_and_reflectivity(
list_of_input_matrices=list_of_cam_matrices,
training_option_dict=training_option_dict)
list_of_cam_matrices[0] = list_of_cam_matrices[0][..., 0]
list_of_cam_matrices[1] = list_of_cam_matrices[1][..., 0]
list_of_guided_cam_matrices = trainval_io.separate_shear_and_reflectivity(
list_of_input_matrices=list_of_guided_cam_matrices,
training_option_dict=training_option_dict)
print('Denormalizing predictors...')
list_of_input_matrices = trainval_io.separate_shear_and_reflectivity(
list_of_input_matrices=list_of_input_matrices,
training_option_dict=training_option_dict)
list_of_input_matrices = model_interpretation.denormalize_data(
list_of_input_matrices=list_of_input_matrices,
model_metadata_dict=model_metadata_dict)
print('Writing class-activation maps to file: "{0:s}"...'.format(
output_file_name))
gradcam.write_standard_file(
pickle_file_name=output_file_name,
list_of_input_matrices=list_of_input_matrices,
list_of_cam_matrices=list_of_cam_matrices,
list_of_guided_cam_matrices=list_of_guided_cam_matrices,
model_file_name=model_file_name,
full_id_strings=full_id_strings,
storm_times_unix_sec=storm_times_unix_sec,
target_class=target_class,
target_layer_name=target_layer_name,
sounding_pressure_matrix_pascals=sounding_pressure_matrix_pascals)
def _run(model_file_name, init_function_name, storm_metafile_name,
num_examples, top_example_dir_name, component_type_string,
target_class, layer_name, neuron_indices, channel_index,
num_iterations, ideal_activation, learning_rate, l2_weight,
radar_constraint_weight, minmax_constraint_weight, output_file_name):
"""Runs backwards optimization on a trained CNN.
This is effectively the main method.
:param model_file_name: See documentation at top of file.
:param init_function_name: Same.
:param storm_metafile_name: Same.
:param num_examples: Same.
:param top_example_dir_name: Same.
:param component_type_string: Same.
:param target_class: Same.
:param layer_name: Same.
:param neuron_indices: Same.
:param channel_index: Same.
:param num_iterations: Same.
:param ideal_activation: Same.
:param learning_rate: Same.
:param l2_weight: Same.
:param radar_constraint_weight: Same.
:param minmax_constraint_weight: Same.
:param output_file_name: Same.
"""
if l2_weight <= 0:
l2_weight = None
if radar_constraint_weight <= 0:
radar_constraint_weight = None
if minmax_constraint_weight <= 0:
minmax_constraint_weight = None
if ideal_activation <= 0:
ideal_activation = None
if init_function_name in ['', 'None']:
init_function_name = None
model_interpretation.check_component_type(component_type_string)
model_metafile_name = '{0:s}/model_metadata.p'.format(
os.path.split(model_file_name)[0])
print(
'Reading model metadata from: "{0:s}"...'.format(model_metafile_name))
model_metadata_dict = cnn.read_model_metadata(model_metafile_name)
input_matrices = None
init_function = None
full_storm_id_strings = None
storm_times_unix_sec = None
sounding_pressure_matrix_pa = None
if init_function_name is None:
print('Reading storm metadata from: "{0:s}"...'.format(
storm_metafile_name))
full_storm_id_strings, storm_times_unix_sec = (
tracking_io.read_ids_and_times(storm_metafile_name))
if 0 < num_examples < len(full_storm_id_strings):
full_storm_id_strings = full_storm_id_strings[:num_examples]
storm_times_unix_sec = storm_times_unix_sec[:num_examples]
example_dict = testing_io.read_predictors_specific_examples(
top_example_dir_name=top_example_dir_name,
desired_full_id_strings=full_storm_id_strings,
desired_times_unix_sec=storm_times_unix_sec,
option_dict=model_metadata_dict[cnn.TRAINING_OPTION_DICT_KEY],
layer_operation_dicts=model_metadata_dict[
cnn.LAYER_OPERATIONS_KEY])
print(SEPARATOR_STRING)
input_matrices = example_dict[testing_io.INPUT_MATRICES_KEY]
sounding_pressure_matrix_pa = example_dict[
testing_io.SOUNDING_PRESSURES_KEY]
num_examples = input_matrices[0].shape[0]
else:
num_examples = 1
init_function = _create_initializer(
init_function_name=init_function_name,
model_metadata_dict=model_metadata_dict)
print('Reading model from: "{0:s}"...'.format(model_file_name))
model_object = cnn.read_model(model_file_name)
output_matrices = None
initial_activations = numpy.full(num_examples, numpy.nan)
final_activations = numpy.full(num_examples, numpy.nan)
for i in range(num_examples):
if init_function_name is None:
this_init_arg = [a[[i], ...] for a in input_matrices]
else:
this_init_arg = init_function
if component_type_string == CLASS_COMPONENT_TYPE_STRING:
print((
'\nOptimizing {0:d}th of {1:d} images for target class {2:d}...'
).format(i + 1, num_examples, target_class))
this_result_dict = backwards_opt.optimize_input_for_class(
model_object=model_object,
target_class=target_class,
init_function_or_matrices=this_init_arg,
num_iterations=num_iterations,
learning_rate=learning_rate,
l2_weight=l2_weight,
radar_constraint_weight=radar_constraint_weight,
minmax_constraint_weight=minmax_constraint_weight,
model_metadata_dict=model_metadata_dict)
elif component_type_string == NEURON_COMPONENT_TYPE_STRING:
print((
'\nOptimizing {0:d}th of {1:d} images for neuron {2:s} in layer'
' "{3:s}"...').format(i + 1, num_examples, str(neuron_indices),
layer_name))
this_result_dict = backwards_opt.optimize_input_for_neuron(
model_object=model_object,
layer_name=layer_name,
neuron_indices=neuron_indices,
init_function_or_matrices=this_init_arg,
num_iterations=num_iterations,
learning_rate=learning_rate,
l2_weight=l2_weight,
ideal_activation=ideal_activation,
radar_constraint_weight=radar_constraint_weight,
minmax_constraint_weight=minmax_constraint_weight,
model_metadata_dict=model_metadata_dict)
else:
print(('\nOptimizing {0:d}th of {1:d} images for channel {2:d} in '
'layer "{3:s}"...').format(i + 1, num_examples,
channel_index, layer_name))
this_result_dict = backwards_opt.optimize_input_for_channel(
model_object=model_object,
layer_name=layer_name,
channel_index=channel_index,
init_function_or_matrices=this_init_arg,
stat_function_for_neuron_activations=K.max,
num_iterations=num_iterations,
learning_rate=learning_rate,
l2_weight=l2_weight,
ideal_activation=ideal_activation,
radar_constraint_weight=radar_constraint_weight,
minmax_constraint_weight=minmax_constraint_weight,
model_metadata_dict=model_metadata_dict)
initial_activations[i] = this_result_dict[
backwards_opt.INITIAL_ACTIVATION_KEY]
final_activations[i] = this_result_dict[
backwards_opt.FINAL_ACTIVATION_KEY]
these_output_matrices = this_result_dict[
backwards_opt.NORM_OUTPUT_MATRICES_KEY]
if output_matrices is None:
output_matrices = [None] * len(these_output_matrices)
for k in range(len(output_matrices)):
if output_matrices[k] is None:
output_matrices[k] = these_output_matrices[k] + 0.
else:
output_matrices[k] = numpy.concatenate(
(output_matrices[k], these_output_matrices[k]), axis=0)
if init_function_name is None:
continue
these_input_matrices = this_result_dict[
backwards_opt.NORM_INPUT_MATRICES_KEY]
if input_matrices is None:
input_matrices = [None] * len(these_input_matrices)
for k in range(len(input_matrices)):
if input_matrices[k] is None:
input_matrices[k] = these_input_matrices[k] + 0.
else:
input_matrices[k] = numpy.concatenate(
(input_matrices[k], these_input_matrices[k]), axis=0)
print(SEPARATOR_STRING)
training_option_dict = model_metadata_dict[cnn.TRAINING_OPTION_DICT_KEY]
print('Denormalizing input examples...')
input_matrices = trainval_io.separate_shear_and_reflectivity(
list_of_input_matrices=input_matrices,
training_option_dict=training_option_dict)
input_matrices = model_interpretation.denormalize_data(
list_of_input_matrices=input_matrices,
model_metadata_dict=model_metadata_dict)
print('Denormalizing optimized examples...')
output_matrices = trainval_io.separate_shear_and_reflectivity(
list_of_input_matrices=output_matrices,
training_option_dict=training_option_dict)
output_matrices = model_interpretation.denormalize_data(
list_of_input_matrices=output_matrices,
model_metadata_dict=model_metadata_dict)
print('Writing results to: "{0:s}"...'.format(output_file_name))
bwo_metadata_dict = backwards_opt.check_metadata(
component_type_string=component_type_string,
num_iterations=num_iterations,
learning_rate=learning_rate,
target_class=target_class,
layer_name=layer_name,
ideal_activation=ideal_activation,
neuron_indices=neuron_indices,
channel_index=channel_index,
l2_weight=l2_weight,
radar_constraint_weight=radar_constraint_weight,
minmax_constraint_weight=minmax_constraint_weight)
backwards_opt.write_standard_file(
pickle_file_name=output_file_name,
denorm_input_matrices=input_matrices,
denorm_output_matrices=output_matrices,
initial_activations=initial_activations,
final_activations=final_activations,
model_file_name=model_file_name,
metadata_dict=bwo_metadata_dict,
full_storm_id_strings=full_storm_id_strings,
storm_times_unix_sec=storm_times_unix_sec,
sounding_pressure_matrix_pa=sounding_pressure_matrix_pa)
def _apply_upconvnet_one_file(example_file_name, num_examples,
upconvnet_model_object, cnn_model_object,
cnn_metadata_dict, cnn_feature_layer_name,
upconvnet_file_name, top_output_dir_name):
"""Applies upconvnet to examples from one file.
:param example_file_name: Path to input file (will be read by
`input_examples.read_example_file`).
:param num_examples: Number of examples to read.
:param upconvnet_model_object: Trained upconvnet (instance of
`keras.models.Model` or `keras.models.Sequential`).
:param cnn_model_object: Trained CNN (instance of
`keras.models.Model` or `keras.models.Sequential`).
:param cnn_metadata_dict: Dictionary returned by `cnn.read_model_metadata`.
param cnn_feature_layer_name: Name of CNN layer whose output is the feature
vector, which is the input to the upconvnet.
:param upconvnet_file_name: See documentation at top of file.
:param top_output_dir_name: Same.
"""
# Do housekeeping.
training_option_dict = cnn_metadata_dict[cnn.TRAINING_OPTION_DICT_KEY]
training_option_dict[trainval_io.EXAMPLE_FILES_KEY] = [example_file_name]
if cnn_metadata_dict[cnn.LAYER_OPERATIONS_KEY] is not None:
generator_object = testing_io.gridrad_generator_2d_reduced(
option_dict=training_option_dict,
desired_num_examples=num_examples,
list_of_operation_dicts=cnn_metadata_dict[
cnn.LAYER_OPERATIONS_KEY])
elif cnn_metadata_dict[cnn.CONV_2D3D_KEY]:
generator_object = testing_io.myrorss_generator_2d3d(
option_dict=training_option_dict,
desired_num_examples=num_examples)
else:
generator_object = testing_io.generator_2d_or_3d(
option_dict=training_option_dict,
desired_num_examples=num_examples)
# Apply upconvnet.
full_storm_id_strings = []
storm_times_unix_sec = numpy.array([], dtype=int)
reconstructed_radar_matrix = None
mse_by_example = numpy.array([], dtype=float)
while True:
try:
this_storm_object_dict = next(generator_object)
print('\n')
except StopIteration:
break
full_storm_id_strings += this_storm_object_dict[
testing_io.FULL_IDS_KEY]
storm_times_unix_sec = numpy.concatenate(
(storm_times_unix_sec,
this_storm_object_dict[testing_io.STORM_TIMES_KEY]))
these_input_matrices = this_storm_object_dict[
testing_io.INPUT_MATRICES_KEY]
this_actual_matrix = these_input_matrices[0]
this_reconstructed_matrix = upconvnet.apply_upconvnet(
cnn_input_matrices=these_input_matrices,
cnn_model_object=cnn_model_object,
cnn_feature_layer_name=cnn_feature_layer_name,
ucn_model_object=upconvnet_model_object,
verbose=True)
print(MINOR_SEPARATOR_STRING)
if reconstructed_radar_matrix is None:
reconstructed_radar_matrix = this_reconstructed_matrix + 0.
else:
reconstructed_radar_matrix = numpy.concatenate(
(reconstructed_radar_matrix, this_reconstructed_matrix),
axis=0)
num_dimensions = len(this_actual_matrix.shape)
all_axes_except_first = numpy.linspace(1,
num_dimensions - 1,
num=num_dimensions - 1,
dtype=int).tolist()
these_mse = numpy.mean(
(this_actual_matrix - this_reconstructed_matrix)**2,
axis=tuple(all_axes_except_first))
mse_by_example = numpy.concatenate((mse_by_example, these_mse))
print(MINOR_SEPARATOR_STRING)
if len(full_storm_id_strings) == 0:
return
print('Mean sqaured error = {0:.3e}'.format(numpy.mean(mse_by_example)))
# Denormalize reconstructed images.
print('Denormalizing reconstructed radar images...')
metadata_dict_no_soundings = copy.deepcopy(cnn_metadata_dict)
metadata_dict_no_soundings[cnn.TRAINING_OPTION_DICT_KEY][
trainval_io.SOUNDING_FIELDS_KEY] = None
option_dict_no_soundings = metadata_dict_no_soundings[
cnn.TRAINING_OPTION_DICT_KEY]
denorm_recon_radar_matrices = trainval_io.separate_shear_and_reflectivity(
list_of_input_matrices=[reconstructed_radar_matrix],
training_option_dict=option_dict_no_soundings)
if training_option_dict[trainval_io.UPSAMPLE_REFLECTIVITY_KEY]:
denorm_recon_radar_matrices[0] = trainval_io.downsample_reflectivity(
reflectivity_matrix_dbz=denorm_recon_radar_matrices[0][..., 0])
denorm_recon_radar_matrices[0] = numpy.expand_dims(
denorm_recon_radar_matrices[0], axis=-1)
denorm_recon_radar_matrices = model_interpretation.denormalize_data(
list_of_input_matrices=denorm_recon_radar_matrices,
model_metadata_dict=metadata_dict_no_soundings)
# Write reconstructed images.
spc_date_string = time_conversion.time_to_spc_date_string(
numpy.median(storm_times_unix_sec).astype(int))
output_file_name = upconvnet.find_prediction_file(
top_directory_name=top_output_dir_name,
spc_date_string=spc_date_string,
raise_error_if_missing=False)
print('Writing predictions to: "{0:s}"...'.format(output_file_name))
upconvnet.write_predictions(
pickle_file_name=output_file_name,
denorm_recon_radar_matrices=denorm_recon_radar_matrices,
full_storm_id_strings=full_storm_id_strings,
storm_times_unix_sec=storm_times_unix_sec,
mse_by_example=mse_by_example,
upconvnet_file_name=upconvnet_file_name)