-
Notifications
You must be signed in to change notification settings - Fork 0
/
EEG_Train_add_new_EEG.py
132 lines (100 loc) · 5.15 KB
/
EEG_Train_add_new_EEG.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
#!!!!!
import torch
import torch.nn.functional as F
import matplotlib.pyplot as plt
import random
import numpy as np
####################### 从 EEG_data_A.txt 和 EEG_data_B.txt 读取 EEG数据############################
total_EEG_data_number=1000#读取n条EEG数据
total_EEG_Features=24 #这是固定的。每一条EEG数据都有24个参数值。
training_times=500 #训练的次数
learn_rate=0.02 #学习率
torch.set_default_dtype(torch.float64)
f = open('EEG record\EEG_data_Ruanjiyang.txt', 'r')
All_EEG_data_lines=f.readlines()
EEG_data_A=torch.Tensor(total_EEG_data_number,total_EEG_Features)
for k in range(total_EEG_data_number):
EEG_data_one_line=(All_EEG_data_lines[k].split('A')) ####按照字符"A"来截断每一条EEG数据，分割成24小份
# print(EEG_data_one_line)
for i in range(total_EEG_Features):
EEG_data_A[k][i]=float(EEG_data_one_line[i])
f.close()
###############读取EEG_data_A.txt完毕######################
f = open('EEG record\EEG_data_Ruanjingheng.txt', 'r')
All_EEG_data_lines=f.readlines()
EEG_data_B=torch.Tensor(total_EEG_data_number,total_EEG_Features)
for k in range(total_EEG_data_number):
EEG_data_one_line=(All_EEG_data_lines[k].split('A')) ####按照字符"A"来截断每一条EEG数据，分割成24小份
for i in range(total_EEG_Features):
EEG_data_B[k][i]=float(EEG_data_one_line[i])
f.close()
###############读取EEG_data_B.txt完毕######################
###########################从 OPENBCI读取 EEG数据结束##########################
########################开始生成 数据样本
# n_data = torch.ones(total_EEG_data_number, total_EEG_Features)
# x0 = torch.normal(EEG_data_A, 20) # class0 x data (tensor), shape=(100, 2)
y0 = torch.ones(total_EEG_data_number) # class0 y data (tensor), shape=(100, 1)
# x1 = torch.normal(EEG_data_B, 20) # class1 x data (tensor), shape=(100, 2)
y1 = torch.zeros(total_EEG_data_number) # class1 y data (tensor), shape=(100, 1)
x = torch.cat((EEG_data_A,EEG_data_B), ).type(torch.DoubleTensor) # shape (200, 2) FloatTensor = 32-bit floating
# print(x.size())
y = torch.cat((y0, y1), ).type(torch.LongTensor) # shape (200,) LongTensor = 64-bit integer
################ 数据整理!!!! ###################
for i in range(total_EEG_data_number*2):
x[i][2]=x[i][2]/10
x[i][3]=x[i][3]/10
x[i][4]=x[i][4]/10
x[i][20]=x[i][4]/2
x[i][21]=x[i][4]/2
################数据整理###################
####################### 绘制原始数据 ##########################
###########准备X轴##################
x_zhou=torch.ones(x.size())
for i in range(total_EEG_data_number*2):
for j in range(total_EEG_Features):
x_zhou[i][j]=j+random.randint(-4,4)/10 #加了一些随机量，否则都在一直线上了。
# ####################################
# tensor_2_draw=x ##指定绘制哪个Tensor
# prediction = torch.max(tensor_2_draw, 1)[1]
# pred_y = prediction.data.numpy()
# for i in range(0,23,1):
# plt.scatter(x_zhou.data.numpy()[:, i], tensor_2_draw.data.numpy()[:, i], c=y.data.numpy(), s=5+ tensor_2_draw.data.numpy()[:, i]*4, lw=1, cmap='RdYlGn',alpha=0.2)
# plt.show()
###############################################################
###################转入到 Cuda#####################3
if torch.cuda.is_available():
print("Cuda is installed")
x_zhou=x_zhou.cuda()
x=x.cuda()
y=y.cuda()
print("Change to Cuda")
#################### 开始训练#####################################
net=torch.load('ruanjiyang_net.pkl') #读取训练好的结果
optimizer = torch.optim.SGD(net.parameters(), lr=learn_rate)
loss_func = torch.nn.CrossEntropyLoss().cuda() # the target label is NOT an one-hotted
#loss_func = torch.nn.MSELoss()
plt.ion() # something about plotting
for t in range(training_times):
out = net(x) # input x and predict based on x
loss = loss_func(out, y) # must be (1. nn output, 2. target), the target label is NOT one-hotted
optimizer.zero_grad() # clear gradients for next train
loss.backward() # backpropagation, compute gradients
optimizer.step() # apply gradients
if t % 2 == 0:
# plot and show learning process
plt.cla()
prediction = torch.max(out, 1)[1]
pred_y = prediction.data.cpu().numpy()
target_y = y.data.cpu().numpy()
#plt.plot(x_zhou.data.numpy(), x.data.numpy(), color=pred_y, linewidth=1, alpha=0.6)
for i in range(0,24,1):
plt.scatter(x_zhou.data.cpu().numpy()[:, i], x.data.cpu().numpy()[:, i], c=pred_y, s=0.2+x.data.cpu().numpy()[:, i]*5, lw=1, cmap='RdYlGn',alpha=0.2)
loss_output=loss*100
accuracy = (float((pred_y == target_y).astype(int).sum()) / float(target_y.size))*100
out_text='Accuracy=%.2f'%accuracy +'% '+ 'Loss=%.2f' % loss_output +'%\n'+ 'Learning times=%d ' %t + 'Learning Rate=%.3f' %learn_rate
plt.text(5,np.max(x.data.cpu().numpy())-5, out_text,fontdict={'size': 20, 'color': 'red'})
plt.pause(0.1)
plt.ioff()
plt.show()
# #################### 训练结束 #####################################
torch.save(net, 'ruanjiyang_net.pkl') #保存好训练结果