-
Notifications
You must be signed in to change notification settings - Fork 0
/
init.py
executable file
·128 lines (95 loc) · 3.93 KB
/
init.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
import os
import shutil
from torch.utils import data
import cv2
import torch
import numpy as np
from torchvision import datasets, transforms
import copy
from client import Client
from server import Server
from utils import parse, gif_maker
def dataset_sel(dataset_name):
root = 'datasets/'
return {
"MNIST": datasets.MNIST(root, download=True, train=True, transform=transforms.Compose(
[transforms.ToTensor(), transforms.Normalize([0.5], [0.5])])),
"CIFAR10": datasets.CIFAR10(root, train=True, transform=transforms.Compose(
[transforms.ToTensor(), transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))]
), target_transform=None, download=True) # transforms.Normalize((0.1307,), (0.3081,))
}.get(dataset_name)
def load_datasets(clients, config):
train_dataset = dataset_sel(config.dataset)
num_data_total = int(config.num_data_owned_setup * len(clients))
train_distributer = data.DataLoader(dataset=train_dataset,
batch_size=num_data_total,
shuffle=config.shuffle
)
images, labels = next(iter(train_distributer))
# Heterogeneous
if config.order:
images_array = images.numpy()
labels_array = labels.numpy()
images_list = []
labels_list = []
for i in range(config.n_classes):
for j in range(num_data_total):
if labels_array[j] == i:
images_list.append(images_array[j])
labels_list.append(labels_array[j])
images_array = np.array(images_list)
labels_array = np.array(labels_list)
images = torch.from_numpy(images_array)
labels = torch.from_numpy(labels_array)
else:
images, labels = next(iter(train_distributer))
for client in clients:
for i in range(config.num_data_owned_setup):
j = i + client.id * config.num_data_owned_setup
client.load_data([images[j], labels[j]])
def init_federated():
# clients list
clients = []
# load configs
config = parse()
# generate clients
for i in range(config.num_of_clients):
clients.append(Client(id=i, config=config))
# generate server
server = Server(id_num=0, config=config)
if os.path.exists(server.model_dir + server.generator_name) and os.path.exists(
server.model_dir + server.discriminator_name):
server.load_model()
print("Global model saved on the server has been restored!")
elif not (os.path.exists(server.model_dir + server.generator_name) or os.path.exists(
server.model_dir + server.discriminator_name)):
print("Global model has been created!")
else:
raise EOFError
# load datasets
load_datasets(clients=clients, config=config)
# load models
for client in clients:
client.load_model(generator=copy.deepcopy(server.generator),
discriminator=copy.deepcopy(server.discriminator))
print("Client {}'s model has been updated from the server".format(client.id))
return clients, server, config
if __name__ == '__main__':
clients, server, config = init_federated()
gif_maker(clients=clients, config=config)
"""pic = np.array([])
for i in range(50):
pic_h = np.array([])
for j in range(50):
if j == 0:
pic_h = np.array(clients[9].dataset[i * 50 + j]).transpose(1, 2, 0)
else:
pic_h = np.hstack((pic_h, np.array(clients[9].dataset[i * 50 + j]).transpose(1, 2, 0)))
if i == 0:
pic = pic_h
else:
pic = np.vstack((pic, pic_h))
cv2.imshow("pic", pic)
cv2.waitKey(0)
shutil.rmtree("clients")
# shutil.rmtree("servers")"""