forked from shaoxiongji/federated-learning
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathmain_sota.py
179 lines (160 loc) · 7.77 KB
/
main_sota.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
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# Python version: 3.6
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import copy
import numpy as np
from torchvision import datasets, transforms
import torch
from utils.sampling import mnist_iid, mnist_noniid, cifar_iid
from utils.options import args_parser
from models.Update import LocalUpdate
from models.Nets import MLP, CNNMnist, CNNCifar, ResNetMnist
from models.Fed import FedAvg
from models.test import test_img, test_img_2
import pandas as pd
from sota.sota import Median as Median, TrimmedMean
import os
os.environ['CUDA_VISIBLE_DEVICES'] = '0'
def FedAvg_test(untargeted_rate, random_rate, imbalance_degree, args_dataset, args_model, args, aggre_algo):
args.device = torch.device('cuda:{}'.format(args.gpu) if torch.cuda.is_available() and args.gpu != -1 else 'cpu')
test_accuracy_list = []
# load dataset and split users
if args_dataset == 'mnist':
trans_mnist = transforms.Compose([transforms.ToTensor(), transforms.Normalize((0.1307,), (0.3081,))])
dataset_train = datasets.MNIST('../data/mnist/', train=True, download=True, transform=trans_mnist)
dataset_test = datasets.MNIST('../data/mnist/', train=False, download=True, transform=trans_mnist)
# sample users
if args.iid:
print("choose iid")
dict_users = mnist_iid(dataset_train, args.num_users)
else:
print("choose non iid")
dict_users = mnist_noniid(dataset_train, args.num_users, imbalance_degree)
elif args_dataset == 'fashion':
trans_mnist = transforms.Compose([transforms.ToTensor(), transforms.Normalize((0.1307,), (0.3081,))])
dataset_train = datasets.FashionMNIST('../data/fashion/', train=True, download=True, transform=trans_mnist)
dataset_test = datasets.FashionMNIST('../data/fashion/', train=False, download=True, transform=trans_mnist)
# sample users
if args.iid:
print("choose iid")
dict_users = mnist_iid(dataset_train, args.num_users)
else:
print("choose non iid")
dict_users = mnist_noniid(dataset_train, args.num_users, imbalance_degree)
elif args_dataset == 'cifar':
trans_cifar = transforms.Compose([transforms.ToTensor(), transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])
dataset_train = datasets.CIFAR10('../data/cifar', train=True, download=True, transform=trans_cifar)
dataset_test = datasets.CIFAR10('../data/cifar', train=False, download=True, transform=trans_cifar)
if args.iid:
dict_users = cifar_iid(dataset_train, args.num_users)
else:
exit('Error: only consider IID setting in CIFAR10')
else:
exit('Error: unrecognized dataset')
img_size = dataset_train[0][0].shape
# build model
if args_model == 'cnn' and args_dataset == 'cifar':
net_glob = CNNCifar(args=args).to(args.device)
elif args_model == 'cnn' and args_dataset == 'mnist':
net_glob = CNNMnist(args=args).to(args.device)
elif args_model == 'resnet' and args_dataset == 'mnist':
net_glob = ResNetMnist(args=args).to(args.device)
elif args_model == 'resnet' and args_dataset == 'fashion':
net_glob = ResNetMnist(args=args).to(args.device)
elif args_model == 'mlp':
len_in = 1
for x in img_size:
len_in *= x
net_glob = MLP(dim_in=len_in, dim_hidden=200, dim_out=args.num_classes).to(args.device)
else:
exit('Error: unrecognized model')
net_glob.train()
# copy weights
w_glob = net_glob.state_dict()
# training
loss_train = []
cv_loss, cv_acc = [], []
val_loss_pre, counter = 0, 0
net_best = None
best_loss = None
val_acc_list, net_list = [], []
if args.all_clients:
print("Aggregation over all clients")
w_locals = [w_glob for i in range(args.num_users)]
for iter in range(args.epochs):
loss_locals = []
if not args.all_clients:
w_locals = []
m = max(int(args.frac * args.num_users), 1)
print(m)
untargeted_num = int(untargeted_rate * m)
random_num = int(random_rate * m)
idxs_users = np.random.choice(range(args.num_users), m, replace=False)
for idx in idxs_users:
if(untargeted_num>0):
#print("attack")
local = LocalUpdate(args=args, dataset=dataset_train, idxs=dict_users[idx], isUntargeted=True)
untargeted_num -= 1
elif(random_num>0):
local = LocalUpdate(args=args, dataset=dataset_train, idxs=dict_users[idx], isRandom=True)
random_num -= 1
else:
local = LocalUpdate(args=args, dataset=dataset_train, idxs=dict_users[idx])
w, loss = local.train(net=copy.deepcopy(net_glob).to(args.device))
if args.all_clients:
w_locals[idx] = copy.deepcopy(w)
else:
w_locals.append(copy.deepcopy(w))
loss_locals.append(copy.deepcopy(loss))
# 根据对framework的选择，调用Median、Trimmed Mean或者FedAvg聚合函数，聚合local updates成global updates
if(aggre_algo=="Median"):
w_glob = Median(w_locals)
elif(aggre_algo=="TrimmedMean"):
w_glob = TrimmedMean(w_locals)
elif(aggre_algo=="FedAvg"):
w_glob = FedAvg(w_locals)
# 将global updates后的weight复制给net_glob，即全局网络模型
net_glob.load_state_dict(w_glob)
# testing 增加在每一轮对global model的test
net_glob.eval()
acc_train, loss_train = test_img(net_glob, dataset_train, args)
acc_test, loss_test = test_img(net_glob, dataset_test, args)
print("Round {:3d},Training accuracy: {:.2f}".format(iter, acc_train))
print("Testing accuracy: {:.2f}".format(acc_test))
test_accuracy_list.append(acc_test)
return test_accuracy_list
def exp_main(folder_addr, attack_rate_list, imbalance_degree_list, attack_type="random", dataset_type='mnist', model_type = 'resnet', framework_type = "Median"):
args = args_parser()
if(attack_type == 'untargeted'):
random_rate = 0.0
for untargeted_rate in attack_rate_list:
temp_accuracy_list = []
for imbalance_degree in imbalance_degree_list:
accuracy_list = FedAvg_test(untargeted_rate, random_rate, imbalance_degree, dataset_type, model_type, args, framework_type)
temp_accuracy_list.append(accuracy_list)
temp_accuracy_dataframe = pd.DataFrame(temp_accuracy_list)
temp_accuracy_dataframe.to_csv(folder_addr+"/"+framework_type+"/"+attack_type+"/"+dataset_type+"/"+str(untargeted_rate)+".csv")
if (attack_type == 'random'):
untargeted_rate_rate = 0.0
for random_rate in attack_rate_list:
temp_accuracy_list = []
for imbalance_degree in imbalance_degree_list:
accuracy_list = FedAvg_test(untargeted_rate, random_rate, imbalance_degree, dataset_type, model_type,
args, framework_type)
temp_accuracy_list.append(accuracy_list)
temp_accuracy_dataframe = pd.DataFrame(temp_accuracy_list)
temp_accuracy_dataframe.to_csv(
folder_addr + "/" + framework_type + "/" + attack_type + "/" + dataset_type + "/" + str(
random_rate) + ".csv")
if __name__ == '__main__':
folder_addr = "exp_data"
attack_rate_list = [0.1,0.2,0.4,0.5]
imbalance_degree_list = [0.1,0.2,0.4,0.5]
attack_type = 'untargeted'
dataset_type = 'mnist'
model_type = 'resnet'
framework_type = "Median"
exp_main(folder_addr, attack_rate_list, imbalance_degree_list, attack_type, dataset_type, model_type, framework_type)