-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathsampling.py
251 lines (225 loc) · 9.47 KB
/
sampling.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
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# Python version: 3.6
import numpy as np
import copy
import itertools
from random import shuffle
from torchvision import datasets, transforms
# np.random.seed(1)
def unique_index(L, f):
return [i for (i, value) in enumerate(L) if value == f]
def mnist_iid(args, dataset, num_users, num_items):
"""
Sample I.I.D. client data from MNIST dataset
:param dataset:
:param num_users:
:return: dict of image index
"""
dict_users, all_idxs = {}, [i for i in range(len(dataset))]
if len(dataset) == 60000:
if args.iid == True:
num_digits = int(num_items / 10)
labels = dataset.train_labels.numpy()
classes = np.unique(labels)
classes_index = []
for i in range(len(classes)):
classes_index.append(unique_index(labels, classes[i]))
for i in range(num_users):
c = []
for j in range(10):
b = (np.random.choice(classes_index[j], num_digits, \
replace=False))
for m in range(num_digits):
c.append(b[m])
# print(c)
dict_users[i] = set(c)
else:
dict_users, all_idxs = {}, [i for i in range(len(dataset))]
for i in range(num_users):
dict_users[i] = set(np.random.choice(all_idxs, num_items, replace=False))
if num_users * num_items <= len(dataset):
all_idxs = list(set(all_idxs) - dict_users[i])
else:
c = set(np.random.choice(all_idxs, num_items, replace=False))
for i in range(num_users):
dict_users[i] = copy.deepcopy(c)
# print("\nDivide", len(all_idxs))
return dict_users
def mnist_noniid(args, dataset, num_users, num_items):
"""
Sample non-I.I.D client data from MNIST dataset
:param dataset:
:param num_users:
:return:
"""
if args.dataset == 'mnist':
# divide and assign
num_digit_noniid = 4
dict_users = {}
labels = dataset.train_labels.numpy()
classes = np.unique(labels)
classes_index = []
for i in range(len(classes)):
classes_index.append(unique_index(labels, classes[i]))
digit_ch_list = list(itertools.combinations(range(len(classes)), num_digit_noniid))
digit_ch_idx = [i for i in range(len(digit_ch_list))]
shuffle(digit_ch_idx)
digit_ch_list_stor = copy.deepcopy(digit_ch_list)
num_items_iid = int(np.ceil((1 - args.degree_noniid) * num_items / len(classes)))
num_items_noniid = int(np.ceil(args.degree_noniid * num_items / num_digit_noniid))
k = 0
for i in digit_ch_idx:
digit_ch_list[i] = copy.deepcopy(digit_ch_list_stor[k])
k += 1
for i in range(num_users):
c = []
for j in range(len(classes)):
b = (np.random.choice(classes_index[j], int(num_items_iid), \
replace=False))
classes_index[j] = list(set(classes_index[j]) - set(b))
for m in range(num_items_iid):
c.append(b[m])
for j in list(digit_ch_list[i]):
b = (np.random.choice(classes_index[j], int(num_items_noniid), \
replace=False))
classes_index[j] = list(set(classes_index[j]) - set(b))
for m in range(num_items_noniid):
c.append(b[m])
dict_users[i] = set(c)
else:
dict_users, all_idxs = {}, [i for i in range(len(dataset))]
c = set(np.random.choice(all_idxs, num_items, replace=False))
for i in range(num_users):
dict_users[i] = copy.deepcopy(c)
# if num_users*num_items <= len(dataset):
# all_idxs = list(set(all_idxs) - dict_users[i])
return dict_users
def FashionMNIST_iid(args, dataset, num_users, num_items):
"""
Sample I.I.D. client data from FashionMNIST dataset
:param dataset:
:param num_users:
:return: dict of image index
"""
dict_users, all_idxs = {}, [i for i in range(len(dataset))]
if len(dataset) == 60000:
if args.strict_iid == True:
num_digits = int(num_items / 10)
labels = dataset.train_labels.numpy()
classes = np.unique(labels)
classes_index = []
for i in range(len(classes)):
classes_index.append(unique_index(labels, classes[i]))
for i in range(num_users):
c = []
for j in range(10):
b = (np.random.choice(classes_index[j], num_digits, \
replace=False))
for m in range(num_digits):
c.append(b[m])
# print(c)
dict_users[i] = set(c)
else:
dict_users, all_idxs = {}, [i for i in range(len(dataset))]
for i in range(num_users):
dict_users[i] = set(np.random.choice(all_idxs, num_items, replace=False))
if num_users * num_items <= len(dataset):
all_idxs = list(set(all_idxs) - dict_users[i])
else:
c = set(np.random.choice(all_idxs, num_items, replace=False))
for i in range(num_users):
dict_users[i] = copy.deepcopy(c)
# print("\nDivide", len(all_idxs))
return dict_users
def FashionMNIST_noniid(args, dataset, num_users, num_items):
"""
Sample non-I.I.D client data from FashionMNIST dataset
:param dataset:
:param num_users:
:return:
"""
if args.dataset == 'FashionMNIST':
# divide and assign
num_digit_noniid = 4
dict_users = {}
labels = dataset.train_labels.numpy()
classes = np.unique(labels)
classes_index = []
for i in range(len(classes)):
classes_index.append(unique_index(labels, classes[i]))
digit_ch_list = list(itertools.combinations(range(len(classes)), num_digit_noniid))
digit_ch_idx = [i for i in range(len(digit_ch_list))]
shuffle(digit_ch_idx)
digit_ch_list_stor = copy.deepcopy(digit_ch_list)
num_items_iid = int(np.ceil((1 - args.degree_noniid) * num_items / len(classes)))
num_items_noniid = int(np.ceil(args.degree_noniid * num_items / num_digit_noniid))
k = 0
for i in digit_ch_idx:
digit_ch_list[i] = copy.deepcopy(digit_ch_list_stor[k])
k += 1
for i in range(num_users):
c = []
for j in range(len(classes)):
b = (np.random.choice(classes_index[j], int(num_items_iid), \
replace=False))
classes_index[j] = list(set(classes_index[j]) - set(b))
for m in range(num_items_iid):
c.append(b[m])
for j in list(digit_ch_list[i]):
b = (np.random.choice(classes_index[j], int(num_items_noniid), \
replace=False))
classes_index[j] = list(set(classes_index[j]) - set(b))
for m in range(num_items_noniid):
c.append(b[m])
dict_users[i] = set(c)
else:
dict_users, all_idxs = {}, [i for i in range(len(dataset))]
c = set(np.random.choice(all_idxs, num_items, replace=False))
for i in range(num_users):
dict_users[i] = copy.deepcopy(c)
# if num_users*num_items <= len(dataset):
# all_idxs = list(set(all_idxs) - dict_users[i])
return dict_users
def cifar_iid(dataset, num_users):
"""
Sample I.I.D. client data from CIFAR10 dataset
:param dataset:
:param num_users:
:return: dict of image index
"""
num_items = int(len(dataset) / num_users)
dict_users, all_idxs = {}, [i for i in range(len(dataset))]
for i in range(num_users):
dict_users[i] = set(np.random.choice(all_idxs, num_items, replace=False))
all_idxs = list(set(all_idxs) - dict_users[i])
return dict_users
# Divide into 100 portions of total data. Allocate 2 random portions for each user
def cifar_noniid(dataset, num_users):
num_shards, num_imgs = 100, 500
idx_shard = [i for i in range(num_shards)]
dict_users = {i: np.array([]) for i in range(num_users)}
idxs = np.arange(num_shards * num_imgs)
labels = np.array(dataset.train_labels) # .numpy()
print(len(idxs))
print(len(labels))
# sort labels
idxs_labels = np.vstack((idxs, labels))
idxs_labels = idxs_labels[:, idxs_labels[1, :].argsort()]
idxs = idxs_labels[0, :]
# divide and assign
for i in range(num_users):
rand_set = set(np.random.choice(idx_shard, 4, replace=False))
# idx_shard = list(set(idx_shard) - rand_set)
for rand in rand_set:
dict_users[i] = np.concatenate((dict_users[i], idxs[rand * num_imgs:(rand + 1) * num_imgs]), axis=0)
# np.random.shuffle(dict_users[i])
return dict_users
if __name__ == '__main__':
dataset_train = datasets.MNIST('./data/FashionMNIST/', train=True, download=True,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307,), (0.3081,))
]))
num = 100
d = mnist_noniid(dataset_train, num)