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(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 check_metadata(component_type_string,
num_iterations,
learning_rate,
target_class=None,
layer_name=None,
ideal_activation=None,
neuron_indices=None,
channel_index=None,
l2_weight=None,
radar_constraint_weight=None,
minmax_constraint_weight=None):
"""Error-checks metadata.
:param component_type_string: See doc for `saliency_maps.check_metadata`.
:param num_iterations: Number of iterations.
:param learning_rate: Learning rate.
:param target_class: See doc for `saliency_maps.check_metadata`.
:param layer_name: Same.
:param ideal_activation: Same.
:param neuron_indices: Same.
:param channel_index: Same.
:param l2_weight: Weight for L_2 regularization.
:param radar_constraint_weight: Weight used to multiply part of loss
function with radar constraints (see doc for
`_radar_constraints_to_loss_fn`).
:param minmax_constraint_weight: Weight used to multiply part of loss
function with min-max constraints (see doc for
`_minmax_constraints_to_loss_fn`).
:return: metadata_dict: Dictionary with the following keys.
metadata_dict['component_type_string']: See input doc.
metadata_dict['num_iterations']: Same.
metadata_dict['learning_rate']: Same.
metadata_dict['target_class']: Same.
metadata_dict['layer_name']: Same.
metadata_dict['ideal_activation']: Same.
metadata_dict['neuron_indices']: Same.
metadata_dict['channel_index']: Same.
metadata_dict['l2_weight']: Same.
metadata_dict['radar_constraint_weight']: Same.
metadata_dict['minmax_constraint_weight']: Same.
"""
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)
error_checking.assert_is_integer(num_iterations)
error_checking.assert_is_greater(num_iterations, 0)
error_checking.assert_is_greater(learning_rate, 0.)
error_checking.assert_is_less_than(learning_rate, 1.)
if l2_weight is not None:
error_checking.assert_is_greater(l2_weight, 0.)
if radar_constraint_weight is not None:
error_checking.assert_is_greater(radar_constraint_weight, 0.)
if minmax_constraint_weight is not None:
error_checking.assert_is_greater(minmax_constraint_weight, 0.)
metadata_dict.update({
NUM_ITERATIONS_KEY: num_iterations,
LEARNING_RATE_KEY: learning_rate,
L2_WEIGHT_KEY: l2_weight,
RADAR_CONSTRAINT_WEIGHT_KEY: radar_constraint_weight,
MINMAX_CONSTRAINT_WEIGHT_KEY: minmax_constraint_weight
})
return metadata_dict
def write_file(pickle_file_name,
normalized_predictor_matrix,
saliency_matrix,
model_file_name,
component_type_string,
target_class=None,
layer_name=None,
ideal_activation=None,
neuron_indices=None,
channel_index=None):
"""Writes saliency maps to Pickle file.
E = number of examples
M = number of rows in each grid
N = number of columns in each grid
C = number of channels (predictor variables)
:param pickle_file_name: Path to output file.
:param normalized_predictor_matrix: E-by-M-by-N-by-C numpy array of
normalized predictor values (input images).
:param saliency_matrix: E-by-M-by-N-by-C numpy array of saliency values.
:param model_file_name: Path to file containing trained CNN on which
saliency maps are based. Should be readable by
`traditional_cnn.read_keras_model`.
:param component_type_string: See doc for
`gewittergefahr.deep_learning.saliency_maps.check_metadata`.
:param target_class: Same.
:param layer_name: Same.
:param ideal_activation: Same.
:param neuron_indices: Same.
:param channel_index: Same.
"""
gg_saliency.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)
error_checking.assert_is_numpy_array_without_nan(
normalized_predictor_matrix)
error_checking.assert_is_numpy_array(normalized_predictor_matrix,
num_dimensions=4)
error_checking.assert_is_numpy_array_without_nan(saliency_matrix)
error_checking.assert_is_numpy_array(
saliency_matrix,
exact_dimensions=numpy.array(normalized_predictor_matrix.shape))
error_checking.assert_is_string(model_file_name)
metadata_dict = {
MODEL_FILE_NAME_KEY: model_file_name,
COMPONENT_TYPE_KEY: component_type_string,
TARGET_CLASS_KEY: target_class,
LAYER_NAME_KEY: layer_name,
IDEAL_ACTIVATION_KEY: ideal_activation,
NEURON_INDICES_KEY: neuron_indices,
CHANNEL_INDEX_KEY: channel_index
}
file_system_utils.mkdir_recursive_if_necessary(file_name=pickle_file_name)
pickle_file_handle = open(pickle_file_name, 'wb')
pickle.dump(normalized_predictor_matrix, pickle_file_handle)
pickle.dump(saliency_matrix, pickle_file_handle)
pickle.dump(metadata_dict, pickle_file_handle)
pickle_file_handle.close()
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, 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 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
print 'Reading storm metadata from: "{0:s}"...'.format(storm_metafile_name)
storm_ids, storm_times_unix_sec = tracking_io.read_ids_and_times(
storm_metafile_name)
print SEPARATOR_STRING
if 0 < num_examples < len(storm_ids):
storm_ids = storm_ids[: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_storm_ids=storm_ids,
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
if component_type_string == CLASS_COMPONENT_TYPE_STRING:
print 'Computing saliency maps for target class {0:d}...'.format(
target_class)
list_of_saliency_matrices = (
saliency_maps.get_saliency_maps_for_class_activation(
model_object=model_object,
target_class=target_class,
list_of_input_matrices=list_of_input_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)
list_of_saliency_matrices = (
saliency_maps.get_saliency_maps_for_neuron_activation(
model_object=model_object,
layer_name=layer_name,
neuron_indices=neuron_indices,
list_of_input_matrices=list_of_input_matrices,
ideal_activation=ideal_activation))
else:
print('Computing saliency maps for channel {0:d} in layer "{1:s}"...'
).format(channel_index, layer_name)
list_of_saliency_matrices = (
saliency_maps.get_saliency_maps_for_channel_activation(
model_object=model_object,
layer_name=layer_name,
channel_index=channel_index,
list_of_input_matrices=list_of_input_matrices,
stat_function_for_neuron_activations=K.max,
ideal_activation=ideal_activation))
print 'Denormalizing model inputs...'
list_of_input_matrices = model_interpretation.denormalize_data(
list_of_input_matrices=list_of_input_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=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=output_file_name,
list_of_input_matrices=list_of_input_matrices,
list_of_saliency_matrices=list_of_saliency_matrices,
storm_ids=storm_ids,
storm_times_unix_sec=storm_times_unix_sec,
model_file_name=model_file_name,
saliency_metadata_dict=saliency_metadata_dict,
sounding_pressure_matrix_pascals=sounding_pressure_matrix_pascals)