def _get_num_dataset_refs():
script_path = os.path.dirname(__file__)
datasets = utils.read_all_datasets(script_path, 'TSC')
dataset_classes = {
dataset_name: y_train.shape[0]
for dataset_name, (_, y_train, _, _) in datasets.items()
}
return dataset_classes
root_dir = config['PATH']['root_directory']
# xps are hyperparameters.
xps = [
'use_bottleneck', 'use_residual', 'nb_filters', 'depth', 'kernel_size',
'batch_size'
]
if sys.argv[1] == 'InceptionTime':
# run nb_iter_ iterations of Inception on the whole archive indicated in config file
classifier_name = 'inception' # train the model first
archive_name = config['ARCHIVE']['archive_name']
nb_iter_ = config['INCEPTION_MODEL']['Train'][
'num_of_Inception_modules'] # number of Inception modules in the network
datasets_dict = read_all_datasets(root_dir, 1)
for iter in range(nb_iter_):
print('\t\titer', iter)
trr = ''
if iter != 0:
trr = '_itr_' + str(iter)
tmp_output_directory = root_dir + '/results/' + classifier_name + '/' + archive_name + trr + '/'
for dataset_name in config['ARCHIVE']['data_name_list']:
print('\t\t\tdataset_name: ', dataset_name)
x_train, y_train, x_test, y_test, y_true, nb_classes, y_true_train, enc = prepare_data(
)
root_dir = '/mnt/nfs/casimir/'
results_dir = root_dir + 'results/fcn/'
batch_size = 16
nb_epochs = 10
verbose = False
write_dir_root = root_dir + '/transfer-learning-results/'
if sys.argv[1] == 'transfer_learning':
# loop through all archives
for archive_name in ARCHIVE_NAMES:
# read all datasets
datasets_dict = read_all_datasets(root_dir, archive_name)
# loop through all datasets
for dataset_name in ALL_DATASET_NAMES:
# get the directory of the model for this current dataset_name
output_dir = results_dir + archive_name + '/' + dataset_name + '/'
# loop through all the datasets to transfer to the learning
for dataset_name_tranfer in ALL_DATASET_NAMES:
# check if its the same dataset
if dataset_name == dataset_name_tranfer:
continue
# set the output directory to write new transfer learning results
write_output_dir = write_dir_root+archive_name+'/'+dataset_name+\
'/'+dataset_name_tranfer+'/'
write_output_dir = create_directory(write_output_dir)
if write_output_dir is None:
continue
do_ensemble = True
if do_ensemble:
root_dir_output = root_deep_learning_dir + 'results/ensemble/'
else:
if do_data_augmentation:
root_dir_output = root_deep_learning_dir + 'results/resnet_augment/'
else:
root_dir_output = root_deep_learning_dir + 'results/resnet/'
from resnet import Classifier_RESNET
# loop the archive names
for archive_name in ARCHIVE_NAMES:
# read all the datasets
datasets_dict = read_all_datasets(root_dir_dataset_archive, archive_name)
# loop through all the dataset names
for dataset_name in DATASET_NAMES:
print('dataset_name: ', dataset_name)
# read dataset
x_train, y_train, x_test, y_test, nb_classes, classes, max_prototypes, \
init_clusters = read_data_from_dataset(use_init_clusters=False)
# specify the output directory for this experiment
output_dir = root_dir_output + archive_name + '/' + dataset_name + '/'
_, classes_counts = np.unique(y_train, return_counts=True)
# this means that all classes will have a number of time series equal to
# nb_prototypes
nb_prototypes = classes_counts.max()
}
classifier_obj = classifier_factory.get_classifier(classifier_name)
classifier = classifier_obj(output_directory, classifier_params)
classifier.fit(x_train, y_train, x_test, y_test, y_true)
# main
if sys.argv[1] == 'run_all':
for classifier_name in constants.CLASSIFIERS:
print('classifier_name', classifier_name)
for archive_name in constants.ARCHIVE_NAMES:
print('\tarchive_name', archive_name)
datasets_dict = utils.read_all_datasets(ROOT_DIR, archive_name)
for _iter in range(constants.ITERATIONS):
print('\t\titer : {}'.format(_iter))
trr = ''
if _iter != 0:
trr = '_itr_{}'.format(_iter)
tmp_output_directory = os.path.join(ROOT_DIR, 'results',
classifier_name,
archive_name + trr)
for dataset_name in constants.dataset_names_for_archive[
archive_name]:
print('\t\t\tdataset_name: ', dataset_name)
def main():
start = time.time()
args = parser.parse_args()
if args.approach == 1:
print('Conduct evaluation using approach 1.')
epochs = ''
if args.iter is not None:
iterations = args.iter
else:
iterations = ITERATIONS
if args.cls_epochs is not None:
epochs = args.cls_epochs
if args.cls is not None:
if args.cls == 'allCls':
for classifier_name in CLASSIFIERS:
ARCHIVE_NAME = ARCHIVE_NAMES[0]
#print('\tarchive_name', ARCHIVE_NAME)
if epochs == '':
epochs = CLS_EPOCHS[classifier_name]
if args.es_patience is None:
cl = classifier_name + '_ep' + str(epochs)
else:
cl = classifier_name + '_ep' + str(
epochs) + '_patience' + str(args.es_patience)
print('\tclassifier_name', cl)
datasets_dict = read_all_datasets(ROOT_DIR, ARCHIVE_NAME)
tmp_output_directory = ROOT_DIR + '/results/' + cl + '/approach1_iter' + str(
iterations) + '/'
if args.aug == 'noAug':
augmentator = None
augmentator_name = 'NoAug'
run_iterations(args, augmentator, augmentator_name,
tmp_output_directory, iterations,
datasets_dict, classifier_name, epochs,
start)
if args.aug == 'allAug':
for aug in AUG:
if aug == 'Random':
for parameters in RANDOM_AUG_PARAMETERS:
diagonal = parameters[0]
down = parameters[1]
right = parameters[2]
each = parameters[3]
n = parameters[4]
augmentator = RandomAug.RandomAug(
n, diagonal, down, right, each)
augmentator_name = aug + '_diag' + str(
diagonal) + '_down' + str(
down) + '_right' + str(
right) + '_each' + str(
int(each)) + '_n' + str(n)
run_iterations(args, augmentator,
augmentator_name,
tmp_output_directory,
iterations, datasets_dict,
classifier_name, epochs,
start)
if aug == 'PRA':
for n in PRA_N:
augmentator = PartlyRandomAug.PartlyRandomAug(
n)
augmentator_name = aug + '_n' + str(n)
run_iterations(args, augmentator,
augmentator_name,
tmp_output_directory,
iterations, datasets_dict,
classifier_name, epochs,
start)
if aug == 'Per':
for n in PER_N:
for prob in PER_PROB:
augmentator = PermutationAug.PermutationAug(
n, prob)
augmentator_name = aug + '_n' + str(
n) + '_prob' + str(prob)
run_iterations(args, augmentator,
augmentator_name,
tmp_output_directory,
iterations,
datasets_dict,
classifier_name, epochs,
start)
if args.approach == 2:
print('Conduct evaluation using approach 2.')
epochs = ''
if args.cv is not None:
cv = args.cv
else:
cv = CV
if args.cls_epochs is not None:
epochs = args.cls_epochs
if args.cls is not None:
if args.cls == 'allCls':
for classifier_name in CLASSIFIERS:
ARCHIVE_NAME = ARCHIVE_NAMES[0]
if epochs == '':
epochs = CLS_EPOCHS[classifier_name]
if args.es_patience is None:
cl = classifier_name + '_ep' + str(epochs)
else:
cl = classifier_name + '_ep' + str(
epochs) + '_patience' + str(args.es_patience)
print('\tclassifier_name', cl)
datasets_dict = read_all_datasets(ROOT_DIR, ARCHIVE_NAME)
tmp_output_directory = ROOT_DIR + '/results/' + cl + '/approach2_cv' + str(
cv) + '/'
if args.aug == 'noAug':
augmentator = None
augmentator_name = 'NoAug'
run_cv(args, augmentator, augmentator_name,
tmp_output_directory, datasets_dict,
classifier_name, epochs, start, cv)
if args.aug == 'allAug':
for aug in AUG:
if aug == 'Random':
for parameters in RANDOM_AUG_PARAMETERS:
diagonal = parameters[0]
down = parameters[1]
right = parameters[2]
each = parameters[3]
n = parameters[4]
augmentator = RandomAug.RandomAug(
n, diagonal, down, right, each)
augmentator_name = aug + '_diag' + str(
diagonal) + '_down' + str(
down) + '_right' + str(
right) + '_each' + str(
int(each)) + '_n' + str(n)
run_cv(args, augmentator, augmentator_name,
tmp_output_directory, datasets_dict,
classifier_name, epochs, start, cv)
if aug == 'PRA':
for n in PRA_N:
augmentator = PartlyRandomAug.PartlyRandomAug(
n)
augmentator_name = aug + '_n' + str(n)
run_cv(args, augmentator, augmentator_name,
tmp_output_directory, datasets_dict,
classifier_name, epochs, start, cv)
if aug == 'Per':
for n in PER_N:
for prob in PER_PROB:
augmentator = PermutationAug.PermutationAug(
n, prob)
augmentator_name = aug + '_n' + str(
n) + '_prob' + str(prob)
run_cv(args, augmentator,
augmentator_name,
tmp_output_directory,
datasets_dict, classifier_name,
epochs, start, cv)
if args.generate_results_overview:
generate_results_overview()
copy_eval_results_and_check_best_models()
print('Results overview generated.')
if args.plot_epochs_overview:
plot_epochs_overview()
print('Epochs overview plotted.')