-
Notifications
You must be signed in to change notification settings - Fork 0
/
az_p3_learning.py
278 lines (258 loc) · 13.1 KB
/
az_p3_learning.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
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
import numpy as np
import scipy.signal as ss
import matplotlib.pyplot as plt
from class_SysParam import SystemParameters
from library.mean_and_std_of_imgs import mean_of_imgs, std_of_imgs
from library.define_folder_name import my_folder_name
from library.evaluate_models import evaluate_model_and_get_decoded_imgs
from library.save_and_load_AE_model import save_my_history, save_my_model
from lib_detection_method.metrics_WRITING_PAPER import compute_TP_and_FN_given_NonAttack
from lib_detection_method.metrics_WRITING_PAPER import compute_FP_and_TN_given_Attack
# from lib_detection_method.metrics_WRITING_PAPER import compute_FP_and_TN_with_SSD
import os
os.environ["CUDA_VISIBLE_DEVICES"] = "-1" # use CPU
""" Load the system parameters """
SysParam = SystemParameters()
n_Tx = SysParam.n_Tx # number of transmitters
n_Rx = SysParam.n_Rx # number of receive antennas
snrdB_Bob = SysParam.snrdB_Bob # in dB
snrdB_Eve = SysParam.snrdB_Eve # in dB
DOA_Bob_list = SysParam.DOA_Bob_list # in degrees
DOA_Eve = SysParam.DOA_Eve # in degrees
K = SysParam.Rician_factor
NLOS = SysParam.n_NLOS_paths
""" Name the folder that contains the data """
folder_name = my_folder_name(n_Tx, n_Rx,
snrdB_Bob, DOA_Bob_list,
snrdB_Eve, DOA_Eve,
K, NLOS)
cur_path = os.path.abspath(os.getcwd())
path_to_datasets = os.path.join(cur_path, 'input/' + folder_name)
path_to_results = os.path.join(cur_path, 'results/' + folder_name)
""" Check if a subfolder exists """
if not os.path.exists(path_to_results): # check if the folder exists
os.mkdir(path_to_results)
os.mkdir(path_to_results + '/PF') # create the folder
# ============================================================================
""" Load the training and testing datasets """
imgs_train_normal = np.load(path_to_datasets + '/imgs_train_normal.npy')
imgs_test_normal = np.load(path_to_datasets + '/imgs_test_normal.npy')
DOA_Eve_changes = DOA_Eve
folder_name__DOA_Eve_changes = my_folder_name(n_Tx, n_Rx,
snrdB_Bob, DOA_Bob_list,
snrdB_Eve, DOA_Eve_changes,
K, NLOS)
path_to_imgs_anomalous = os.path.join(cur_path, 'input/' + folder_name__DOA_Eve_changes)
imgs_test_anomalous = np.load(path_to_imgs_anomalous + '/imgs_test_anomalous.npy')
# imgs_test = np.vstack((imgs_test_normal, imgs_test_anomalous))
# ============================================================================
labels_test_normal = np.zeros([len(imgs_test_normal), 1])
labels_test_anomalous = np.ones([len(imgs_test_anomalous), 1])
labels_test = np.vstack((labels_test_normal, labels_test_anomalous))
# ============================================================================
num_angles = 180
angles = np.linspace(-num_angles/2, num_angles/2-1, num_angles)
# ============================================================================
""" Metrics versus priority factor """
def metrics_wrt_priority_factor(imgs_test_normal,
imgs_test_normal_decoded,
imgs_test_anomalous,
imgs_test_anomalous_decoded,
DOA_Bob_list,
avg__diff_NORMAL,
standard_SSD):
scaling_factor_given_H0 = 10.0
scaling_factor_given_H1 = 4.0
acc_list = []
TPR_list = []
FNR_list = []
TNR_list = []
FPR_list = []
priority_factor_list = [0.1*i for i in range(11)]
for priority_factor in priority_factor_list:
TPos, FNega = compute_TP_and_FN_given_NonAttack(imgs_test_normal,
imgs_test_normal_decoded,
DOA_Bob_list,
avg__diff_NORMAL,
standard_SSD,
scaling_factor_given_H0,
priority_factor)
###
FPos, TNega = compute_FP_and_TN_given_Attack(imgs_test_anomalous,
imgs_test_anomalous_decoded,
DOA_Bob_list,
avg__diff_NORMAL,
standard_SSD,
scaling_factor_given_H1,
priority_factor)
###
acc = (TPos + TNega)/(TPos + TNega + FPos + FNega)
acc_list.append(acc)
###
TPR = TPos/(TPos + FNega)
FNR = FNega/(TPos + FNega)
TPR_list.append(TPR)
FNR_list.append(FNR)
###
TNR = TNega/(TNega + FPos)
FPR = FPos/(FPos + TNega)
TNR_list.append(TNR)
FPR_list.append(FPR)
return acc_list, TPR_list, FNR_list, TNR_list, FPR_list
# ============================================================================
""" Metrics versus scaling factor """
def metrics_wrt_scaling_factor(imgs_test_normal,
imgs_test_normal_decoded,
imgs_test_anomalous,
imgs_test_anomalous_decoded,
DOA_Bob_list,
avg__diff_NORMAL,
standard_SSD):
priority_factor__fixed = 1.0
acc_list = []
TPR_list = []
FNR_list = []
TNR_list = []
FPR_list = []
scaling_factor_list = [1*i for i in range(11)]
for scaling_factor in scaling_factor_list:
TPos, FNega = compute_TP_and_FN_given_NonAttack(imgs_test_normal,
imgs_test_normal_decoded,
DOA_Bob_list,
avg__diff_NORMAL,
standard_SSD,
scaling_factor,
priority_factor__fixed)
###
FPos, TNega = compute_FP_and_TN_given_Attack(imgs_test_anomalous,
imgs_test_anomalous_decoded,
DOA_Bob_list,
avg__diff_NORMAL,
standard_SSD,
scaling_factor,
priority_factor__fixed)
###
acc = (TPos + TNega)/(TPos + TNega + FPos + FNega)
acc_list.append(acc)
###
TPR = TPos/(TPos + FNega)
FNR = FNega/(TPos + FNega)
TPR_list.append(TPR)
FNR_list.append(FNR)
###
TNR = TNega/(TNega + FPos)
FPR = FPos/(FPos + TNega)
TNR_list.append(TNR)
FPR_list.append(FPR)
return acc_list, TPR_list, FNR_list, TNR_list, FPR_list
# ============================================================================
n_repeats = 1
n_epochs = 50
priority_factor_list = [0.1*i for i in range(11)]
scaling_factor_list = [0.1*i for i in range(11)]
#
loss_train__multiple_runs = np.zeros([n_repeats, n_epochs])
loss_test__multiple_runs = np.zeros([n_repeats, n_epochs])
#
avg_acc_vs_priority_factor = np.zeros(len(priority_factor_list))
avg_acc_vs_scaling_factor = np.zeros(len(scaling_factor_list))
# metrics versus priority_factor
avg_TPR_vs_priority_factor = np.zeros(len(priority_factor_list))
avg_FNR_vs_priority_factor = np.zeros(len(priority_factor_list))
avg_TNR_vs_priority_factor = np.zeros(len(priority_factor_list))
avg_FPR_vs_priority_factor = np.zeros(len(priority_factor_list))
# metrics versus scaling_factor
avg_TPR_vs_scaling_factor = np.zeros(len(scaling_factor_list))
avg_FNR_vs_scaling_factor = np.zeros(len(scaling_factor_list))
avg_TNR_vs_scaling_factor = np.zeros(len(scaling_factor_list))
avg_FPR_vs_scaling_factor = np.zeros(len(scaling_factor_list))
#
avg_labels_test_decoded = np.zeros([len(labels_test), 1])
for i in range(n_repeats):
# evaluate model
model, history, loss_train, loss_test, \
imgs_train_normal_decoded, \
imgs_test_normal_decoded, \
imgs_test_anomalous_decoded \
= evaluate_model_and_get_decoded_imgs(imgs_train_normal,
imgs_test_normal,
imgs_test_anomalous,
num_angles,
n_epochs)
# Saving only one model is enough for plotting the loss-vs-epoch
if i == 0:
""" Save the history as a dictionary for later use """
save_my_history(history, folder_name + '/PF')
""" Save the trained AUTO-ENCODER model """
save_my_model(model, folder_name + '/PF')
###
""" Calculate the standard-SSD based on ...
... imgs_train_normal and imgs_train_normal_decoded """
diff_NORMAL_DecodedNORMAL = imgs_train_normal - imgs_train_normal_decoded
avg__diff_NORMAL = mean_of_imgs(diff_NORMAL_DecodedNORMAL)
std__diff_NORMAL = std_of_imgs(diff_NORMAL_DecodedNORMAL)
standard_SSD = np.linalg.norm(std__diff_NORMAL)**2
""" Metrics versus priority factor """
acc_vs_priority_factor, TPR_vs_priority_factor, FNR_vs_priority_factor, \
TNR_vs_priority_factor, FPR_vs_priority_factor \
= metrics_wrt_priority_factor(imgs_test_normal,
imgs_test_normal_decoded,
imgs_test_anomalous,
imgs_test_anomalous_decoded,
DOA_Bob_list,
avg__diff_NORMAL,
standard_SSD)
""" Metrics versus scaling factor """
acc_vs_scaling_factor, TPR_vs_scaling_factor, FNR_vs_scaling_factor, \
TNR_vs_scaling_factor, FPR_vs_scaling_factor \
= metrics_wrt_scaling_factor(imgs_test_normal,
imgs_test_normal_decoded,
imgs_test_anomalous,
imgs_test_anomalous_decoded,
DOA_Bob_list,
avg__diff_NORMAL,
standard_SSD)
# end of for loop
avg_acc_vs_priority_factor += np.array(acc_vs_priority_factor)/n_repeats
avg_acc_vs_scaling_factor += np.array(acc_vs_scaling_factor)/n_repeats
#
avg_TPR_vs_priority_factor += np.array(TPR_vs_priority_factor)/n_repeats
avg_TPR_vs_scaling_factor += np.array(TPR_vs_scaling_factor)/n_repeats
avg_FNR_vs_priority_factor += np.array(FNR_vs_priority_factor)/n_repeats
avg_FNR_vs_scaling_factor += np.array(FNR_vs_scaling_factor)/n_repeats
#
avg_TNR_vs_priority_factor += np.array(TNR_vs_priority_factor)/n_repeats
avg_TNR_vs_scaling_factor += np.array(TNR_vs_scaling_factor)/n_repeats
avg_FPR_vs_priority_factor += np.array(FPR_vs_priority_factor)/n_repeats
avg_FPR_vs_scaling_factor += np.array(FPR_vs_scaling_factor)/n_repeats
# end of for loop
# ============================================================================
""" Save avg_labels_test_decoded """
np.save(cur_path + '/results/' + folder_name
+ '/PF/avg_labels_test_decoded.npy',
avg_labels_test_decoded)
""" Save accuracy_versus_alpha, TPR_versus_alpha, TNR_versus_alpha """
np.save(cur_path + '/results/' + folder_name + '/PF/avg_acc_vs_priority.npy',
avg_acc_vs_priority_factor)
np.save(cur_path + '/results/' + folder_name + '/PF/avg_acc_vs_scaling.npy',
avg_acc_vs_scaling_factor)
#
np.save(cur_path + '/results/' + folder_name + '/PF/avg_TPR_vs_priority.npy',
avg_TPR_vs_priority_factor)
np.save(cur_path + '/results/' + folder_name + '/PF/avg_TPR_vs_scaling.npy',
avg_TPR_vs_scaling_factor)
#
np.save(cur_path + '/results/' + folder_name + '/PF/avg_TNR_vs_priority.npy',
avg_TNR_vs_priority_factor)
np.save(cur_path + '/results/' + folder_name + '/PF/avg_TNR_vs_scaling.npy',
avg_TNR_vs_scaling_factor)
#
np.save(cur_path + '/results/' + folder_name + '/PF/avg_FPR_vs_priority.npy',
avg_FPR_vs_priority_factor)
np.save(cur_path + '/results/' + folder_name + '/PF/avg_FPR_vs_scaling.npy',
avg_FPR_vs_scaling_factor)
#
np.save(cur_path + '/results/' + folder_name + '/PF/avg_FNR_vs_priority.npy',
avg_FNR_vs_priority_factor)
np.save(cur_path + '/results/' + folder_name + '/PF/avg_FNR_vs_scaling.npy',
avg_FNR_vs_scaling_factor)