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
|
"""
Date: 2022-03-22
Author: [email protected]
Desc: plot
"""
import matplotlib.pyplot as plt
from matplotlib.lines import Line2D
import numpy as np
def plt_f1():
families = [
'Dridex',
'Gozi',
'Quakbot',
'Tofsee'
]
x = np.array([0,1,2,3])
total_width, n = 0.8, 4
width = total_width / n
x = x - width - width
fsnet = [0.9947,1.,0.9844,0.9842]
autoender = [0.995, 0.96, 0.9448, 0.9201]
fsnet_after_attack = [0.7779, 0.7333, 0.6736, 0.6423]
autoender_after_attack = [0.5285, 0.3322, 0.6570, 0.4857]
plt.bar(x, fsnet, width=width, label='target model')
plt.bar(x+width, autoender, width, label='subtitute model')
plt.bar(x+width*2, autoender_after_attack, width, label='subtitute after attack')
plt.bar(x+width*3, fsnet_after_attack, width, label='target after attack')
# plt.xlim([-1, 4])
plt.title('F1 value before/after adversarial sample attack')
plt.xlabel('botnet family')
plt.ylabel('f1')
plt.legend(bbox_to_anchor=(1.05, 1.0), loc='upper left')
plt.tight_layout()
plt.xticks(x+width, families)
plt.show()
def plt_edr():
families = [
'Dridex',
'Gozi',
'Quakbot',
'Tofsee'
]
x = np.array([0,1,2,3])
total_width, n = 0.4, 2
width = total_width / n
fsnet = [0.825, 0.95, 0.58, 0.8]
autoender = [0.885, 1.0, 0.59, 0.845]
plt.bar(x, fsnet, width=width, label='target model')
plt.bar(x+width, autoender, width, label='subtitute model')
plt.ylim([0, 1])
plt.xlabel('botnet family')
plt.ylabel('EDR(Escape Detection Rate)')
plt.title("adversarial samples Escape Detection Rate")
plt.legend(bbox_to_anchor=(1.05, 1.0), loc='upper left')
plt.tight_layout()
plt.xticks(x+width, families)
plt.show()
def plot_edr_gamma():
plt.figure(dpi=100, figsize=(24, 8))
truncate = 32
rnn_fsnet_Tofsee = np.load("../rnn_fsnet_Tofsee_edr_gamma_list.npy")[:truncate]
rnn_fsnet_Dridex = np.load("../rnn_fsnet_Dridex_edr_gamma_list.npy")[:truncate]
rnn_fsnet_quackbot = np.load("../rnn_fsnet_Quakbot_edr_gamma_list.npy")[:truncate]
rnn_fsnet_TrickBot = np.load("../rnn_fsnet_TrickBot_edr_gamma_list.npy")[:truncate]
rnn_fsnet_Gozi = np.load("../rnn_fsnet_Gozi_edr_gamma_list.npy")[:truncate]
maxLength = 1600
plt.plot([x[0]/maxLength for x in rnn_fsnet_Tofsee], [x[1] for x in rnn_fsnet_Tofsee],
marker='s', color='cyan', linestyle='-', markersize=4, linewidth='2')
plt.plot([x[0]/maxLength for x in rnn_fsnet_Tofsee], [x[2] for x in rnn_fsnet_Tofsee],
marker="^", color="cyan", linestyle=':', markersize=6, linewidth='2')
plt.plot([x[0]/maxLength for x in rnn_fsnet_Dridex], [x[1] for x in rnn_fsnet_Dridex],
marker='s', color='purple', linestyle='-', markersize=4, linewidth='2')
plt.plot([x[0]/maxLength for x in rnn_fsnet_Dridex], [x[2] for x in rnn_fsnet_Dridex],
marker="^", color="purple", linestyle=':', markersize=6, linewidth='2')
plt.plot([x[0]/maxLength for x in rnn_fsnet_quackbot], [x[1] for x in rnn_fsnet_quackbot],
marker='s', color='#00B04E', linestyle='-', markersize=4, linewidth='2')
plt.plot([x[0]/maxLength for x in rnn_fsnet_quackbot], [x[2] for x in rnn_fsnet_quackbot],
marker="^", color="#00B04E", linestyle=':', markersize=6, linewidth='2')
plt.plot([x[0]/maxLength for x in rnn_fsnet_TrickBot], [x[1] for x in rnn_fsnet_TrickBot],
marker='s', color='#F7921E', linestyle='-', markersize=4, linewidth='2')
plt.plot([x[0]/maxLength for x in rnn_fsnet_TrickBot], [x[2] for x in rnn_fsnet_TrickBot],
marker="^", color="#F7921E", linestyle=':', markersize=6, linewidth='2')
plt.plot([x[0]/maxLength for x in rnn_fsnet_Gozi], [x[1] for x in rnn_fsnet_Gozi],
marker='s', color='red', linestyle='-', markersize=4, linewidth='2')
plt.plot([x[0]/maxLength for x in rnn_fsnet_Gozi], [x[2] for x in rnn_fsnet_Gozi],
marker="^", color="red", linestyle=':', markersize=6, linewidth='2')
labels = [
Line2D([0], [0], color='cyan', lw=4, label="Tofsee"),
Line2D([0], [0], color='purple', lw=4, label="Dridex"),
Line2D([0], [0], color='#00B04E', lw=4, label="Quakbot"),
Line2D([0], [0], color='#F7921E', lw=4, label="TrickBot"),
Line2D([0], [0], color='red', lw=4, label="Gozi"),
Line2D([0], [0], color='black', lw=2, linestyle="-", marker="s", markersize=4, label=r"$\psi_{s}^{rnn}$"),
Line2D([0], [0], color='black', lw=2, linestyle=":", marker="^", markersize=6, label=r"$\psi_{t}^{FS-Net}$"),
]
plt.legend(handles=labels, ncol=7)
x_ticks = np.arange(0, 0.11, 0.01)
plt.xticks(x_ticks)
plt.ylabel("EDR")
plt.xlabel(r"$\epsilon$")
# plt.title(r"$\psi_{s}^{rnn}$-$\psi_{t}^{FS-Net}$")
plt.savefig("../fig/edr_epsilon_a.eps", dpi=200, format='eps', bbox_inches="tight", pad_inches=0)
plt.show()
def plot_edr_targets():
"""
:return:
"""
plt.figure(dpi=100, figsize=(24, 8))
truncate = 100
# todo: 需要补充rnn和autoencoder作为替代模型的迁移攻击下效果
rnn_fsnet_TrickBot = np.load("../rnn_fsnet_TrickBot_edr_gamma_list.npy")[:truncate]
rnn_svm_TrickBot = np.load("../rnn_svm_TrickBot_edr_gamma_list.npy")[:truncate]
rnn_lr_TrickBot = np.load("../rnn_lr_TrickBot_edr_gamma_list.npy")[:truncate]
rnn_dt_TrickBot = np.load("../rnn_dt_TrickBot_edr_gamma_list.npy")[:truncate]
rnn_rf_TrickBot = np.load("../rnn_rf_TrickBot_edr_gamma_list.npy")[:truncate]
rnn_knn_TrickBot = np.load("../rnn_knn_TrickBot_edr_gamma_list.npy")[:truncate]
rnn_dnn_TrickBot = np.load("../rnn_dnn_TrickBot_edr_gamma_list.npy")[:truncate]
rnn_lstm_TrickBot = np.load("../rnn_lstm_TrickBot_edr_gamma_list.npy")[:truncate]
maxLength = 1600
# plt.plot([x[0]/maxLength for x in rnn_fsnet_TrickBot], [x[1] for x in rnn_fsnet_TrickBot], label=r"$\psi_s^{LSTM}$",
# color='red', linestyle=':', linewidth='2')
plt.plot([x[0]/maxLength for x in rnn_fsnet_TrickBot], [x[1] for x in rnn_fsnet_TrickBot], label=r"$\psi_s^{LSTM}$",
marker='^', color='red', linestyle=':', markersize=4, linewidth='2')
plt.plot([x[0]/maxLength for x in rnn_fsnet_TrickBot], [x[2] for x in rnn_fsnet_TrickBot], label=r"$\psi_t^{FS-Net}$",
color="blue", linestyle='-', linewidth='2', marker='s', markersize=4)
plt.plot([x[0]/maxLength for x in rnn_dnn_TrickBot], [x[1] for x in rnn_dnn_TrickBot], label=r"$\psi_t^{DNN}$",
color='brown', linestyle='-', linewidth='2', marker='s', markersize=4)
plt.plot([x[0]/maxLength for x in rnn_lstm_TrickBot], [x[1] for x in rnn_lstm_TrickBot], label=r"$\psi_t^{LSTM}$",
color='olive', linestyle='-', linewidth='1', marker='s', markersize=4)
plt.plot([x[0]/maxLength for x in rnn_svm_TrickBot], [x[1] for x in rnn_svm_TrickBot], label=r"$\psi_t^{SVM}$",
marker='s', color='green', linestyle='-', markersize=4, linewidth='2')
plt.plot([x[0]/maxLength for x in rnn_lr_TrickBot], [x[1] for x in rnn_lr_TrickBot], label=r"$\psi_t^{LR}$",
marker='s', color='yellow', linestyle='-', markersize=4, linewidth='2')
plt.plot([x[0]/maxLength for x in rnn_dt_TrickBot], [x[1] for x in rnn_dt_TrickBot], label=r"$\psi_t^{DT}$",
marker='s', color='cyan', linestyle='-', markersize=4, linewidth='2')
plt.plot([x[0]/maxLength for x in rnn_rf_TrickBot], [x[1] for x in rnn_rf_TrickBot], label=r"$\psi_t^{RF}$",
marker='s', color='magenta', linestyle='-', markersize=4, linewidth='2')
plt.plot([x[0]/maxLength for x in rnn_knn_TrickBot], [x[1] for x in rnn_knn_TrickBot], label=r"$\psi_t^{KNN}$",
marker='s', color='purple', linestyle='-', markersize=4, linewidth='2')
plt.axvline(0.1)
x_ticks = np.arange(0,0.33, 0.02)
plt.xticks(x_ticks)
y_ticks = np.arange(0,1.1,0.1)
plt.yticks(y_ticks)
plt.legend(ncol=9, loc=[0.28,1.001])
# todo: 需要仔细商榷
# plt.title("")
plt.xlabel(r"$\epsilon$")
plt.ylabel(r"EDR")
plt.savefig("../fig/edr_epsilon_b.eps", dpi=200, format='eps', bbox_inches="tight", pad_inches=0)
plt.show()
if __name__ == '__main__':
# plt_edr()
# plt_f1()
plot_edr_gamma()
# plot_edr_targets()
|
