/
SPDNet_Local_Learning.py
172 lines (134 loc) · 6.56 KB
/
SPDNet_Local_Learning.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
##################################################################################################
# FTL Draft Code for Subject-local Analysis
# Author：Ce Ju, Dashan Gao
# Date : July 29, 2020
# Paper : Ce Ju et al., Federated Transfer Learning for EEG Signal Classification, IEEE EMBS 2020.
# Description: One subject(participant) locally train an SPDNetwork for EEG signal classification using its own data.
##################################################################################################
import warnings
import datetime
import numpy as np
import torch
from torch.autograd import Variable
import torch.nn.functional as F
from mne.decoding import CSP
# pyriemann import
from pyriemann.classification import MDM, TSclassifier, FgMDM
from sklearn.pipeline import Pipeline
from sklearn.linear_model import LogisticRegression
from sklearn import svm
warnings.filterwarnings('ignore')
def SPD_experients(cov_data, labels):
import SPDNet
random_index = np.arange(cov_data.shape[0])
np.random.shuffle(random_index)
cov_data = cov_data[random_index, :, :]
labels = labels[random_index]
split_num = int(np.floor(cov_data.shape[0] * 0.8))
cov_data_train = cov_data[0:split_num, :, :]
cov_data_test = cov_data[split_num:cov_data.shape[0], :, :]
print('split_num: ', split_num)
print('rest_num: ', labels.shape[0] - split_num)
print('-------------------------------------------------------')
input_data_train = Variable(torch.from_numpy(cov_data_train)).double()
input_data_test = Variable(torch.from_numpy(cov_data_test)).double()
target_train = Variable(torch.LongTensor(labels[0:split_num]))
target_test = Variable(torch.LongTensor(labels[split_num:labels.shape[0]]))
model = SPDNet.SPDNetwork_2()
for _ in range(500):
stime = datetime.datetime.now()
logits = model(input_data_train)
output = F.log_softmax(logits, dim=-1)
loss = F.nll_loss(output, target_train)
pred = output.data.max(1, keepdim=True)[1]
correct = pred.eq(target_train.data.view_as(pred)).long().cpu().sum()
loss.backward()
lr = 0.1
model.update_all_layers(lr)
etime = datetime.datetime.now()
dtime = etime.second - stime.second
logits = model(input_data_test)
output = F.log_softmax(logits, dim=-1)
pred = output.data.max(1, keepdim=True)[1]
correct_test = pred.eq(target_test.data.view_as(pred)).long().cpu().sum()
return correct_test.item() / pred.shape[0]
if __name__ == '__main__':
# Load data.
data = np.load('raw_data/normalized_original_train_sample.npy')
epoch_data_train = np.load('raw_data/normalized_original_epoch_data_train.npy')
label = np.load('raw_data/train_label.npy')
index = np.load('raw_data/index.npy')
FULL_MDM = []
FULL_FGMDM = []
FULL_TSC = []
FULL_CSP_lr = []
FULL_CSP_svm = []
for bad_subject_index in range(108):
# bad_subject_index = [2, 8, 16, 17, 22, 23, 27, 35, 37, 38, 39, 40, 44, 46, 57, 62, 63, 66, 73, 75, 76, 77, 89, 95, 96, 98, 100, 101]
# bad_subject_index = 2
cov_data_bad = data[bad_subject_index]
labels_bad = label[bad_subject_index]
epochs_data_train_bad = epoch_data_train[bad_subject_index]
MDM_record = []
FGMDM_record = []
TSC_record = []
CSP_lr_record = []
CSP_svm_record = []
for fold in range(1, 6):
train = cov_data_bad[index[bad_subject_index] != fold]
train_CSP = epochs_data_train_bad[index[bad_subject_index] != fold]
train_label = labels_bad[index[bad_subject_index] != fold]
test = cov_data_bad[index[bad_subject_index] == fold]
test_CSP = epochs_data_train_bad[index[bad_subject_index] == fold]
test_label = labels_bad[index[bad_subject_index] == fold]
box_length = np.sum([index[bad_subject_index] == fold])
mdm = MDM(metric=dict(mean='riemann', distance='riemann'))
mdm.fit(train, train_label)
pred = mdm.predict(test)
print('MDM: {:4f}'.format(np.sum(pred == test_label) / box_length))
MDM_record.append(np.sum(pred == test_label) / box_length)
print('-----------------------------------------')
Fgmdm = FgMDM(metric=dict(mean='riemann', distance='riemann'))
Fgmdm.fit(train, train_label)
pred = Fgmdm.predict(test)
print('FGMDM: {:4f}'.format(np.sum(pred == test_label) / box_length))
FGMDM_record.append(np.sum(pred == test_label) / box_length)
print('-----------------------------------------')
clf = TSclassifier()
clf.fit(train, train_label)
pred = clf.predict(test)
print('TSC: {:4f}'.format(np.sum(pred == test_label) / box_length))
TSC_record.append(np.sum(pred == test_label) / box_length)
print('-----------------------------------------')
lr = LogisticRegression()
csp = CSP(n_components=4, reg='ledoit_wolf', log=True)
clf = Pipeline([('CSP', csp), ('LogisticRegression', lr)])
clf.fit(train_CSP, train_label)
pred = clf.predict(test_CSP)
print('CSP_lr: {:4f}'.format(np.sum(pred == test_label) / box_length))
CSP_lr_record.append(np.sum(pred == test_label) / box_length)
print('-----------------------------------------')
lr = svm.SVC(kernel='rbf')
csp = CSP(n_components=4, reg='ledoit_wolf', log=True)
clf = Pipeline([('CSP', csp), ('svc', lr)])
clf.fit(train_CSP, train_label)
pred = clf.predict(test_CSP)
print('CSP_svm: {:4f}'.format(np.sum(pred == test_label) / box_length))
CSP_svm_record.append(np.sum(pred == test_label) / box_length)
print("------------------------------------------------------------------------")
FULL_MDM.append(np.mean(MDM_record))
FULL_TSC.append(np.mean(TSC_record))
FULL_FGMDM.append(np.mean(FGMDM_record))
FULL_CSP_lr.append(np.mean(CSP_lr_record))
FULL_CSP_svm.append(np.mean(CSP_svm_record))
print('MDM Record: ', FULL_MDM)
print('R-Kernel Record: ', FULL_FGMDM)
print('TSC Record: ', FULL_TSC)
print('CSP_lr Record: ', FULL_CSP_lr)
print('CSP_svm Record: ', FULL_CSP_svm)
print('-----------------------')
print('MDM: ', np.mean(FULL_MDM))
print('R-Kernel: ', np.mean(FULL_FGMDM))
print('TSC: ', np.mean(FULL_TSC))
print('CSP_lr: ', np.mean(FULL_CSP_lr))
print('CSP_svm: ', np.mean(FULL_CSP_svm))