forked from muxamilian/privacy-tuw
-
Notifications
You must be signed in to change notification settings - Fork 11
/
auxplot.py
executable file
·171 lines (141 loc) · 6.12 KB
/
auxplot.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
#!/usr/bin/env python3
# An ad-hoc script for plotting tables copy-pasted from LibreOffice Calc
import fileinput
import numpy as np
import matplotlib.pyplot as plt
from matplotlib.patches import Rectangle
import sys
from matplotlib.transforms import Affine2D
from matplotlib.lines import Line2D
group_names = []
values = []
plt.rcParams['font.family'] = 'serif'
plt.rcParams['savefig.format'] = 'pdf'
colors = plt.rcParams['axes.prop_cycle'].by_key()['color']
for line in fileinput.input('-'):
field = line.rstrip().split('\t')
group_names.append(field[0])
values.append(field[1:])
feature_names = values[0]
group_names = group_names[1:]
values = values[1:]
print("feature_names", feature_names)
print("group_names", group_names)
print("values", values)
values = np.array(values, dtype=float)
def importance():
global values
plt.figure(figsize=(5,2.6))
values[values < 0] = 0
values /= np.sum(values,axis=0)
FACTOR = 0.85
# width = 1/(1+values.shape[1])
width = FACTOR / values.shape[1]
order = np.argsort(-np.mean(values, axis=1))
x = np.arange(len(group_names))
for i in range(values.shape[1]):
plt.bar(x + width*i, values[order,i], width, alpha=0.9, label=feature_names[i])
for i in range(values.shape[1]):
plt.bar(x + width*i, np.mean(values[order,:], axis=1), width, color='gray', alpha=0.5, **({'label': 'Mean'} if i==0 else {}), zorder=0)
plt.legend()
plt.xticks(x + width*values.shape[1]/2-width*0.5, [group_names[i] for i in order], rotation=45, horizontalalignment='right')
for tick in plt.gca().xaxis.get_major_ticks():
label = tick.label1
# don't know how to specify translation in axis coordinates, so pdf output
# looks slightly different from plot
label.set_transform(label.get_transform() + Affine2D().translate(5,0))
plt.ylabel('Normalized metric')
plt.tight_layout()
def adv_results():
global values
plt.rcParams['legend.handlelength'] = 1.5
plt.figure(figsize=(5,2.7))
values = np.reshape(values, (-1,values.shape[1]//2,2))
order = (2+np.argsort(np.mean(values[2:,:,1], axis=1))).tolist()
order = [0, 1] + order
width = 0.95 / values.shape[1]
x = np.arange(len(group_names), dtype=float)
x[:2] -= .5
for i in range(values.shape[1]):
plt.bar(x +width*i, values[order,i,1], width, color='gray', alpha=.5, **({'label': 'Unmodified'} if i==0 else {}))
for i in range(values.shape[1]):
plt.bar(x +width*i, values[order,i,0], width, color=colors[i], label=feature_names[i*2].rstrip(' adv'))
plt.legend(loc='lower center', bbox_to_anchor=(0.5,1), ncol=4)
plt.xticks(x + width*values.shape[1]/2-width*0.5, [group_names[i] for i in order], rotation=45, horizontalalignment='right')
for i,label in enumerate(plt.gca().get_xticklabels()):
if i < 2:
label.set_fontweight('bold')
for tick in plt.gca().xaxis.get_major_ticks():
label = tick.label1
label.set_transform(label.get_transform() + Affine2D().translate(8,0))
plt.ylabel('Recall')
plt.tight_layout()
def ars_original():
global group_names
global values
y_labels = [None, "Recall", "Recall", "ARS", "ARS"]
value_indices_to_plot = (1,3)
values = [[float(item) for item in sublist] for sublist in values]
index_by_which_to_sort = 1
old_len = len(group_names)
orig_group_names, orig_values = group_names, values
group_names, values = list(zip(*[(group_name, value) for group_name, value in zip(group_names, values) if value[0] <= 0.5]))
# print(f"Dropped {old_len-len(group_names)} elements.")
robust_groups = set(orig_group_names)-set(group_names)
robust_indices = [orig_group_names.index(item) for item in robust_groups]
print("robust attack types:", "; ".join([f"{g}: orig_flow_recall={v[1]}, adv_flow_recall={v[0]}" for g, v in zip([orig_group_names[index] for index in robust_indices], [orig_values[index] for index in robust_indices])]))
order = list(zip(*sorted(enumerate(values), key=lambda item: item[1][index_by_which_to_sort], reverse=True)))[0]
# print("order", order)
width = 0.75 / len(value_indices_to_plot)
values = np.array(values)
x = np.arange(len(group_names), dtype=float)
fig, ax1 = plt.subplots(figsize=(5,3))
plt.xticks(x + width/2, [group_names[i] for i in order], rotation=45, horizontalalignment='right')
for tick in plt.gca().xaxis.get_major_ticks():
label = tick.label1
label.set_transform(label.get_transform() + Affine2D().translate(8,0))
ax2 = ax1.twinx()
axes = [ax1, ax2]
all_labels = []
for index, i in enumerate(value_indices_to_plot):
label = axes[index].bar(x + width*index, values[order,i], width, color=colors[index], label=feature_names[i])
all_labels.append(label)
axes[index].set_ylabel(y_labels[i])
all_legends = [item.get_label() for item in all_labels]
# ax1.legend(all_legends, all_labels, loc='upper right', bbox_to_anchor=(1,0.95))
# plt.ylim((min(first_quartiles), max(third_quartiles)))
ax2.set_ylabel_legend(Rectangle((0,0), 1, 1, fc=colors[1]), handlelength=1)
ax1.set_ylabel_legend(Rectangle((0,0), 1, 1, fc=colors[0]), handlelength=1)
# ax1.set_ylabel_legend(Rectangle((0,0), 1, 1, fc='gray', alpha=0.2), handlelength=1)
# plt.legend(all_labels, all_legends, loc='lower center', bbox_to_anchor=(0.5,1), ncol=4)
plt.tight_layout()
def ars():
plt.figure(figsize=(5,2))
x_values = [ float(value) for value in group_names ]
plt.plot(x_values, values)
plt.legend(feature_names, ncol=2)
plt.xlabel('Training duration in epochs')
plt.ylabel('ARS')
ylim1,ylim2 = plt.ylim()
plt.ylim((ylim1,ylim2+20)) # move plots away from legend
plt.tight_layout()
def adv():
plt.figure(figsize=(5,2))
x_values = [ float(value) for value in group_names ]
lines = plt.plot(x_values, values[:,:2])
plt.xlabel('Tradeoff $\kappa$')
plt.ylabel('Success ratio')
plt.gca().set_ylabel_legend(Line2D([0],[0], color='gray'), handlelength=1.4)
plt.twinx()
plt.plot(x_values, values[:,2:], linestyle='--')
plt.ylabel('$L_1$ distance')
plt.gca().set_ylabel_legend(Line2D([0],[0], color='gray', linestyle='--'), handlelength=1.4)
plt.legend(lines, ['CIC-IDS-2017', 'UNSW-NB15'])
ylim1,ylim2 = plt.ylim()
plt.ylim((ylim1,11))
# plt.ylim((ylim1,ylim2+20)) # move plots away from legend
plt.tight_layout()
globals()[sys.argv[1]]()
if len(sys.argv) > 2:
plt.savefig(sys.argv[2], bbox_inches = 'tight', pad_inches = 0)
plt.show()