-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathutils.py
193 lines (171 loc) · 8.28 KB
/
utils.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
import sys
import numpy as np
import configparser
import uuid
import json
from tqdm import tqdm
import csv
import os
import random
import math
import time
import subprocess
def loadData(dataset, proportion):
# print("Loading data...\n")
worker_instance_num = {'sentiment': 235, 'rte': 49, 'temp': 61, 'ducks': 108, 'ER': 142, 'SPC': 137}
DATA_PATH = r'./datasets/' + dataset + '/'
ground_truth = np.load(DATA_PATH + 'label_train.npy', allow_pickle=True)
raw_answers = np.load(DATA_PATH + 'data_answer.npy', allow_pickle=True)
raw_ins_num = raw_answers.shape[0]
worker_num = raw_answers.shape[1]
random_missing = random.sample(range(0, raw_ins_num), int(raw_ins_num * (1 - proportion)))
ins_num = raw_ins_num - int(raw_ins_num * (1 - proportion))
answers = np.ones((ins_num, worker_num)) * -1
num = 0
for i in range(raw_ins_num):
if i not in random_missing:
for j in range(worker_num):
answers[num][j] = raw_answers[i][j]
num += 1
answers = answers.astype(np.int32)
normal_workers_aggregation = []
for i in range(ins_num):
votes = np.zeros(2)
for r in range(worker_num):
if answers[i, r] != -1:
votes[answers[i, r]] += 1
normal_workers_aggregation.append(np.argmax(votes))
# print("Loading completed.\n")
return normal_workers_aggregation, ground_truth, answers, raw_ins_num, ins_num, worker_num, random_missing, worker_instance_num[dataset]
def gete2wlandw2el_normal(answers):
e2wl = {}
w2el = {}
label_set=['0','1']
for i in range(answers.shape[0]):
for j in range(answers.shape[1]):
if answers[i][j] != -1:
example, worker, label = str(i),str(j),str(answers[i][j])
if example not in e2wl:
e2wl[example] = []
e2wl[example].append([worker,label])
if worker not in w2el:
w2el[worker] = []
w2el[worker].append([example,label])
return e2wl,w2el,label_set
def gete2wlandw2el(answers, answers_p):
e2wl = {}
w2el = {}
label_set=['0','1']
for i in range(answers.shape[0]):
for j in range(answers.shape[1]):
if answers[i][j] != -1:
example, worker, label = str(i),str(j),str(answers[i][j])
if example not in e2wl:
e2wl[example] = []
e2wl[example].append([worker,label])
if worker not in w2el:
w2el[worker] = []
w2el[worker].append([example,label])
for i in range(answers_p.shape[0]):
for j in range(answers_p.shape[1]):
if answers_p[i][j] != -1:
example, worker, label = str(i),str(j),str(answers_p[i][j])
worker_p = str(int(worker) + answers.shape[1])
if example not in e2wl:
e2wl[example] = []
e2wl[example].append([worker_p,label])
if worker_p not in w2el:
w2el[worker_p] = []
w2el[worker_p].append([example,label])
return e2wl,w2el,label_set
def writeTotalAnswers(answers, dataset, poisonMethod, subMethod, mWorkers, answerPercent, knowPercent, seed):
folder = r'./outputPoisonAnswer/' + dataset
if not os.path.isdir(folder):
os.mkdir(folder)
folder = r'./outputPoisonAnswer/' + dataset + r'/' + poisonMethod
if not os.path.isdir(folder):
os.mkdir(folder)
folder = r'./outputPoisonAnswer/' + dataset + r'/' + poisonMethod + r'/' + subMethod
if not os.path.isdir(folder):
os.mkdir(folder)
poisonDataFile = r'./outputPoisonAnswer/' + dataset + r'/' + poisonMethod + r'/' + subMethod + r'/mW_' + str(mWorkers) + '-answerPercent_' + str(answerPercent) + '-knowPercent_' + str(knowPercent) + '-s_' + str(seed) + '.csv'
with open(poisonDataFile, "w") as w:
writer = csv.writer(w)
writer.writerows([["question", "worker", "answer"]])
[rows, cols] = answers.shape
for row in range(rows):
for col in range(cols):
if (answers[row, col] != -1):
writer.writerows(
[[row, col, answers[row, col]]])
return poisonDataFile
def get_change_rate(truth_file, output_normal, output_normal_and_malicious, method):
# When normal workers make the right questions, the rate at which malicious workers change it
arr0 = [-1] * len(output_normal)
f = open(truth_file, 'r')
reader = csv.reader(f)
next(reader)
for line in reader:
example, label = line
arr0[int(example)] = int(label)
arr1 = [-1] * len(output_normal)
arr2 = [-1] * len(output_normal)
for key in output_normal:
ind = int(key)
if method == "CATD":
arr1[ind] = int(output_normal[key])
else:
if '0' not in output_normal[key]:
output_normal[key]['0'] = 1.0 - output_normal[key]['1']
if '1' not in output_normal[key]:
output_normal[key]['1'] = 1.0 - output_normal[key]['0']
if output_normal[key]['0'] > output_normal[key]['1']:
arr1[ind] = 0
else:
arr1[ind] = 1
for key in output_normal_and_malicious:
ind = int(key)
if method == "CATD":
arr2[ind] = int(output_normal_and_malicious[key])
else:
if '0' not in output_normal_and_malicious[key]:
output_normal_and_malicious[key]['0'] = 1.0 - output_normal_and_malicious[key]['1']
if '1' not in output_normal_and_malicious[key]:
output_normal_and_malicious[key]['1'] = 1.0 - output_normal_and_malicious[key]['0']
if output_normal_and_malicious[key]['0'] > output_normal_and_malicious[
key]['1']:
arr2[ind] = 0
else:
arr2[ind] = 1
numerator = sum(a == b and b != c for a, b, c in zip(arr0, arr1, arr2))
denominator = sum(a == b for a, b in zip(arr0,arr1))
return float(numerator / (denominator * 1.0))
def test(dataset, attackedMethod, poisonDataFile):
normalAnswerFile = "./datasets/" + dataset + "/answer.csv"
output_normal = subprocess.getoutput("python " + r'./attackedMethods/' + attackedMethod + r'/method.py ' + normalAnswerFile + ' ' + '"categorical"').split('\n')[-1]
output_normal_and_malicious = subprocess.getoutput("python " + r'./attackedMethods/' + attackedMethod + r'/method.py ' + poisonDataFile + ' ' + '"categorical"').split('\n')[-1]
truth_file = r"./datasets/" + dataset + r"/truth.csv"
change_rate = get_change_rate(truth_file, eval(output_normal), eval(output_normal_and_malicious),str(attackedMethod))
return change_rate
def run(poisonMethods, subMethods, attackedMethods, datasets, mWorkers, answerPercents, knowPercents, seeds, exp):
for poisonMethod in poisonMethods:
for dataset in datasets:
print()
change_rates = []
for subMethod in subMethods:
for attackedMethod in attackedMethods:
for answerPercent in answerPercents:
for knowPercent in knowPercents:
change_rate = 0
for seed in range(1, seeds + 1):
poisonDataFile = r'./outputPoisonAnswer/' + dataset + r'/' + poisonMethod + r'/' + subMethod + r'/mW_' + str(mWorkers) + '-answerPercent_' + str(answerPercent) + '-knowPercent_' + str(knowPercent) + '-s_' + str(seed) + '.csv'
change_rate += test(dataset, attackedMethod, poisonDataFile)
print(" dataset: %-11s poisonMethod: %-10s subMethod: %-10s attackedMethod: %-10s answerPercent: %.2f knowPercent: %.2f seed: %d change_rate: %f" % (dataset, poisonMethod, subMethod, attackedMethod, answerPercent, knowPercent, seed, change_rate / seeds))
change_rates.append(str(change_rate / seeds))
folder = r'./output/' + exp
if not os.path.isdir(folder):
os.mkdir(folder)
f = open(folder + '/' + 'change_rate_' + dataset + '_' + poisonMethod, 'w')
for change_rate in change_rates:
f.write(change_rate + ',')
f.close()