-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathcompute_cost_sample.py
260 lines (208 loc) · 12.5 KB
/
compute_cost_sample.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
from numpy import disp
import pandas as pd
import numpy as np
from matplotlib import pyplot as plt
import matplotlib.dates as mdates
from data_process import DataPreprocess
time = []
#compute euclidean distance and squared difference
def compute_distance(data, adv_data):
squared_diff = (adv_data-data)**2
sum_squares = pd.DataFrame.sum(pd.DataFrame(squared_diff), axis=1)
return np.sqrt(sum_squares),squared_diff
#load data
def get_data_and_calculate(data, adversarial_data_path, columns, compute_hamming):
adv_data = pd.read_csv(adversarial_data_path)
adv_data = pd.DataFrame(adv_data, columns=columns)
if compute_hamming:
distances, squared_diff = compute_distance(data[columns], adv_data[columns])
#print(np.mean(distances), np.mean(distances[distances>0]))
non_zero = squared_diff[columns].ge(0.0001).sum(axis=1)
print(np.round(np.mean(distances), 3), ' & ',np.round(np.std(distances), 3), ' & ',
len(distances[distances>0.00001]), ' & ', np.round(np.mean(non_zero[non_zero>0].values),3),
' & ',np.round(np.std(non_zero[non_zero>0].values),3),
' & ',np.round(np.max(non_zero[non_zero>0].values),3),
' & ',np.round(np.min(non_zero[non_zero>0].values),3))
else:
distances, _ = compute_distance(data[columns], adv_data[columns])
print(np.round(np.mean(distances), 3), ' & ',np.round(np.std(distances), 3), ' & ',
len(distances[distances>0.0001]))
#load data for SVM
def get_data_and_calculate_SVM(data, adv_data, columns):
distances, squared_diff = compute_distance(pd.DataFrame(data), pd.DataFrame(adv_data))
non_zero = squared_diff.ge(0.0001).sum(axis=1)
print(np.round(np.mean(distances), 3), ' & ',np.round(np.std(distances), 3), ' & ',
len(distances[distances>0.0001]), ' & ', np.round(np.mean(non_zero[non_zero>0].values),3),
' & ',np.round(np.std(non_zero[non_zero>0].values),3),
' & ',np.round(np.max(non_zero[non_zero>0].values),3),
' & ',np.round(np.min(non_zero[non_zero>0].values),3))
#plot visualization
def plotting(distances, ax,maxlim):
ax.plot(range(0,len(distances)), distances)
ax.set_ylim(ymin=0, ymax=maxlim)
ax.set_xlim(xmin=0, xmax=len(distances))
if __name__ == '__main__':
AR = False
LTI = False
PASAD = False
SFIG= True
SVM = False
compute_hamming = LTI or SFIG or SVM
fig, axs = plt.subplots(6,1)
if AR:
data_path = './AR/undersample_test.csv'
data = pd.read_csv(data_path)
time = pd.to_datetime(data['DATETIME'], dayfirst=True)
columns = ['LIT301']
data = pd.DataFrame(data, columns=columns)
print('AR Model replay')
adversaria_data_path = './AR/replay_undersample_jan23.csv'
columns = ['LIT301']
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
print('AR Model random')
adversaria_data_path = './AR/random_replay_undersample_jan23.csv'
columns = ['LIT301']
get_data_and_calculate(data, adversaria_data_path, columns,compute_hamming)
print('AR Model stale')
adversaria_data_path = './AR/stale_undersample_jan23.csv'
columns = ['LIT301']
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
print('AR Model WBC baseline')
adversaria_data_path = './AR/whitebox_attack_all_may22_swat_WBC_baseline.csv'
columns = ['LIT301']
elapsed = './AR/elapsed_vector_WBC_baseline.csv'
elapsed = np.loadtxt(elapsed, delimiter=",")
print(np.mean(elapsed)*1000, np.std(elapsed)*1000)
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
print('AR Model WBC NTP')
adversaria_data_path = './AR/whitebox_attack_all_may22_swat_WBC_NTP.csv'
columns = ['LIT301']
elapsed = './AR/elapsed_vector_WBC_NTP.csv'
elapsed = np.loadtxt(elapsed, delimiter=",")
print(np.mean(elapsed)*1000, np.std(elapsed)*1000)
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
print('AR Model WBC NA')
adversaria_data_path = './AR/whitebox_attack_all_may22_swat_WBC_NA.csv'
columns = ['LIT301']
elapsed = './AR/elapsed_vector_WBC_NA.csv'
elapsed = np.loadtxt(elapsed, delimiter=",")
print(np.mean(elapsed)*1000, np.std(elapsed)*1000)
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
if LTI:
data_path = './Spoofing Framework/SWAT/SWaT_Dataset_Attack_v1.csv'
data = pd.read_csv(data_path)
time = pd.to_datetime(data['Timestamp'], dayfirst=True)
columns = ['FIT101','LIT101','AIT201','AIT202','AIT203','FIT201','DPIT301','FIT301',
'LIT301','AIT401','AIT402','FIT401','LIT401','AIT501','AIT502','AIT503','AIT504',
'FIT501','FIT502','FIT503','FIT504','PIT501','PIT502','PIT503','FIT601']
data = pd.DataFrame(data, columns=columns)
print('LTI Model replay')
adversaria_data_path = './Spoofing Framework/SWAT/unconstrained_spoofing/replay.csv'
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
print('LTI Model random replay')
adversaria_data_path = './Spoofing Framework/SWAT/unconstrained_spoofing/random_replay.csv'
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
print('LTI Model stale')
adversaria_data_path = './Spoofing Framework/SWAT/unconstrained_spoofing/stale.csv'
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
print('LTI Model WBC baseline')
adversaria_data_path = './LTI/whitebox_attack_all_jan23_swat_WBC_baseline.csv'
elapsed = './LTI/elapsed_vector_WBC_baseline_jan23.csv'
elapsed = np.loadtxt(elapsed, delimiter=",")
print(np.mean(elapsed>0)*1000, np.std(elapsed>0)*1000)
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
print('LTI Model WBC NTP')
adversaria_data_path = './LTI/whitebox_attack_all_jan23_swat_WBC_NTP.csv'
elapsed = './LTI/elapsed_vector_WBC_NTP_jan23.csv'
elapsed = np.loadtxt(elapsed, delimiter=",")
print(np.mean(elapsed)*1000,' & ', np.std(elapsed)*1000)
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
print('LTI Model WBC NA')
adversaria_data_path = './LTI/whitebox_attack_all_jan23_swat_WBC_NA.csv'
elapsed = './LTI/elapsed_vector_WBC_NA_jan23.csv'
elapsed = np.loadtxt(elapsed, delimiter=",")
print(np.mean(elapsed)*1000,' & ', np.std(elapsed)*1000)
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
if PASAD:
data_path = './Spoofing Framework/SWAT/SWaT_Dataset_Attack_v1.csv'
data = pd.read_csv(data_path)
import re
time = pd.to_datetime(data['Timestamp'], dayfirst=True)
columns = ['LIT301']
#data = min_max_scaler.fit_transform(data[columns])
data = pd.DataFrame(data, columns=columns)
print('PASAD Model Replay')
adversaria_data_path = './Spoofing Framework/SWAT/unconstrained_spoofing/replay.csv'
columns = ['LIT301']
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
print('PASAD Model Random Replay')
adversaria_data_path = './Spoofing Framework/SWAT/unconstrained_spoofing/random_replay.csv'
columns = ['LIT301']
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
print('PASAD Model Stale')
adversaria_data_path = './Spoofing Framework/SWAT/unconstrained_spoofing/stale.csv'
columns = ['LIT301']
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
print('PASAD Model WBC baseline')
adversaria_data_path = './PASAD/whitebox_attack_all_jan23_swat_WBC_baseline.csv'
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
print('PASAD Model WBC NTP')
adversaria_data_path = './PASAD/whitebox_attack_all_jan23_swat_WBC_NTP.csv'
columns = ['LIT301']
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
print('PASAD Model WBC NA')
adversaria_data_path = './PASAD/whitebox_attack_all_jan23_swat_WBC_NA.csv'
columns = ['LIT301']
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
if SFIG:
data_path = './Spoofing Framework/SWAT/SWaT_Dataset_Attack_v1.csv'
data = pd.read_csv(data_path)
time = pd.to_datetime(data['Timestamp'], dayfirst=True)
columns = ['FIT101','LIT101','MV101','P101','P102','AIT201','AIT202','AIT203','FIT201','MV201','P201','P202','P203','P204','P205','P206','DPIT301','FIT301','LIT301','MV301','MV302','MV303','MV304','P301','P302','AIT401','AIT402','FIT401','LIT401','P401','P402','P403','P404','UV401','AIT501','AIT502','AIT503','AIT504','FIT501','FIT502','FIT503','FIT504','P501','P502','PIT501','PIT502','PIT503','FIT601','P601','P602','P603']
#data = min_max_scaler.fit_transform(data[columns].values)
data = pd.DataFrame(data, columns=columns)
print('SFIG Modelreplay')
adversaria_data_path = './Spoofing Framework/SWAT/unconstrained_spoofing/replay.csv'
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
print('SFIG Model random replay')
adversaria_data_path = './Spoofing Framework/SWAT/unconstrained_spoofing/random_replay.csv'
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
print('SFIG Stale')
adversaria_data_path = './Spoofing Framework/SWAT/unconstrained_spoofing/stale.csv'
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
print('SFIG Model WBC baseline/NTP')
adversaria_data_path = './SFIG/data/whitebox_attack_results_NTP/mod_rows_attack_only_attack_rows.csv'
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
print('SFIG WBC NA')
adversaria_data_path = './SFIG/data/whitebox_attack_results_NA/mod_rows_attack_ALL_new_fix_approx.csv'
get_data_and_calculate(data, adversaria_data_path, columns, compute_hamming)
if SVM:
data_path = './Spoofing Framework/SWAT/SWaT_Dataset_Attack_v1.csv'
time = pd.to_datetime(pd.read_csv(data_path)['Timestamp'], dayfirst=True)
columns = ['LIT101', 'LIT301', 'LIT401']
label_col = ['Normal/Attack']
data = DataPreprocess(data_path, columns, label_col, 1)
print('SVM model replay')
adversaria_data_path = './Spoofing Framework/SWAT/unconstrained_spoofing/replay.csv'
adversaria_data = DataPreprocess(adversaria_data_path, columns, label_col, 1)
get_data_and_calculate_SVM(data.test_data, adversaria_data.test_data, columns)
print('SVM random replay')
adversaria_data_path = './Spoofing Framework/SWAT/unconstrained_spoofing/random_replay.csv'
adversaria_data = DataPreprocess(adversaria_data_path, columns, label_col, 1)
get_data_and_calculate_SVM(data.test_data, adversaria_data.test_data, columns)
print('SVM stale')
adversaria_data_path = './Spoofing Framework/SWAT/unconstrained_spoofing/stale.csv'
adversaria_data = DataPreprocess(adversaria_data_path, columns, label_col, 1)
get_data_and_calculate_SVM(data.test_data, adversaria_data.test_data, columns)
print('SVM WBC baseline')
adversarial_data_path = './SVM/adv_examples_baseline_costrained.npy'
adv_data = np.load(adversarial_data_path, allow_pickle=True)
get_data_and_calculate_SVM(data.test_data, adv_data, columns)
print('SVM WBC NTP')
adversarial_data_path = './SVM/adv_examples_ntp_costrained.npy'
adv_data = np.load(adversarial_data_path, allow_pickle=True)
get_data_and_calculate_SVM(data.test_data, adv_data, columns)
print('SVM WBC NA')
adversarial_data_path = './SVM/adv_examples_na_costrained.npy'
adv_data = np.load(adversarial_data_path, allow_pickle=True)
get_data_and_calculate_SVM(data.test_data, adv_data, columns)