forked from hfawaz/aaltd18
-
Notifications
You must be signed in to change notification settings - Fork 0
/
main.py
executable file
·137 lines (109 loc) · 5.81 KB
/
main.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
from utils.constants import UNIVARIATE_ARCHIVE_NAMES as ARCHIVE_NAMES
from utils.constants import MAX_PROTOTYPES_PER_CLASS
from utils.constants import UNIVARIATE_DATASET_NAMES as DATASET_NAMES
from utils.utils import read_all_datasets
from utils.utils import calculate_metrics
from utils.utils import transform_labels
from utils.utils import create_directory
from utils.utils import plot_pairwise
from augment import augment_train_set
import numpy as np
def augment_function(augment_algorithm_name, x_train, y_train, classes, N, limit_N=True):
if augment_algorithm_name == 'as_dtw_dba_augment':
return augment_train_set(x_train, y_train, classes, N,limit_N = limit_N,
weights_method_name='as', distance_algorithm='dtw'), 'dtw'
def read_data_from_dataset(use_init_clusters=True):
dataset_out_dir = root_dir_output + archive_name + '/' + dataset_name + '/'
x_train = datasets_dict[dataset_name][0]
y_train = datasets_dict[dataset_name][1]
x_test = datasets_dict[dataset_name][2]
y_test = datasets_dict[dataset_name][3]
nb_classes = len(np.unique(np.concatenate((y_train, y_test), axis=0)))
# make the min to zero of labels
y_train, y_test = transform_labels(y_train, y_test)
classes, classes_counts = np.unique(y_train, return_counts=True)
if len(x_train.shape) == 2: # if univariate
# add a dimension to make it multivariate with one dimension
x_train = x_train.reshape((x_train.shape[0], x_train.shape[1], 1))
x_test = x_test.reshape((x_test.shape[0], x_test.shape[1], 1))
# maximum number of prototypes which is the minimum count of a class
max_prototypes = min(classes_counts.max() + 1,
MAX_PROTOTYPES_PER_CLASS + 1)
init_clusters = None
return x_train, y_train, x_test, y_test, nb_classes, classes, max_prototypes, init_clusters
# for mesocentre
##### you should change these for your directories
root_dir = '/b/home/uha/hfawaz-datas/dba-python/'
root_dir_output = root_dir + 'results/'
root_deep_learning_dir = '/b/home/uha/hfawaz-datas/dl-tsc/'
root_dir_dataset_archive = '/b/home/uha/hfawaz-datas/dl-tsc/archives/'
# make sure before doing data augmentation to have the models trained without data augmentation
# in order to use the same weights init method
do_data_augmentation = True
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()
temp = output_dir
# create the directory if not exists
output_dir = create_directory(output_dir)
# check if directory already exists
if output_dir is None:
print('Already_done')
print(temp)
continue
if do_ensemble==False:
# create the resnet classifier
classifier = Classifier_RESNET(output_dir, x_train.shape[1:],
nb_classes, nb_prototypes, classes,
verbose=True, load_init_weights=do_data_augmentation)
if do_data_augmentation:
# augment the dataset
syn_train_set, distance_algorithm = augment_function('as_dtw_dba_augment',
x_train, y_train, classes,
nb_prototypes,limit_N=False)
# get the synthetic train and labels
syn_x_train, syn_y_train = syn_train_set
# concat the synthetic with the reduced random train and labels
aug_x_train = np.array(x_train.tolist() + syn_x_train.tolist())
aug_y_train = np.array(y_train.tolist() + syn_y_train.tolist())
print(np.unique(y_train,return_counts=True))
print(np.unique(aug_y_train,return_counts=True))
y_pred = classifier.fit(aug_x_train, aug_y_train, x_test,
y_test)
else:
# no data augmentation
y_pred = classifier.fit(x_train, y_train, x_test,
y_test)
df_metrics = calculate_metrics(y_test, y_pred, 0.0)
df_metrics.to_csv(output_dir + 'df_metrics.csv', index=False)
print('DONE')
create_directory(output_dir+'DONE')
else:
# for ensemble you will have to compute both models in order to ensemble them
from ensemble import Classifier_ENSEMBLE
classifier_ensemble = Classifier_ENSEMBLE(output_dir, x_train.shape[1:], nb_classes, False)
classifier_ensemble.fit(x_test, y_test)
# plot pairwise once all results are computed for resnet and resnet_augment and ensemble
plot_pairwise(root_deep_learning_dir,root_dir_dataset_archive, 'resnet', 'resnet_augment')