-
Notifications
You must be signed in to change notification settings - Fork 0
/
test_nns.py
270 lines (223 loc) · 11.2 KB
/
test_nns.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
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
import numpy as np
import pandas as pd
import CNN
from sklearn.model_selection import train_test_split
from keras.models import Sequential
from keras.layers import (Dense, Dropout, Conv1D, GlobalAveragePooling1D,
MaxPooling1D, Flatten, Conv2D, Activation,
MaxPooling2D, BatchNormalization, LSTM)
from keras import callbacks
import datetime
import os
from keras.models import Model
from keras import layers
import keras
from keras.wrappers.scikit_learn import KerasClassifier
from keras.utils import np_utils
from sklearn.model_selection import cross_val_score
from sklearn.model_selection import KFold
from sklearn.preprocessing import LabelEncoder
from sklearn.pipeline import Pipeline
num_GPU = 4
epochs = 100
verbose = 1
# 0 | 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 |
# Fp1|Fp2|F7|F3|Fz|F4|F8|FC5|FC1|FC2|FC6|T7|C3|Cz|C4|T8|TP9|CP5|CP1|CP2|CP6|TP10|P7|P3|Pz|P4|P8|PO9|O1|Oz|O2|PO10|
def reshape_1D_conv(X):
X_rashaped = np.array([X[i, :, :].flatten()
for i in range(X.shape[0])]) # concatenate
X_rashaped = X_rashaped.reshape(X_rashaped.shape[0],
X_rashaped.shape[1],
1) # add 3rd dimension
return X_rashaped
def reshape_2D_conv(X):
X_reshaped = X.reshape(X.shape[0], X.shape[1], X.shape[2], 1)
return X_reshaped
# Read data
y = np.load('y.npy')
# ################ TRAIN MODELS #########################
# ConvLSTM2D
# X = np.load('EEG_2D_LSTM.npy')
# X = X.reshape(X.shape[0], X.shape[1], X.shape[2], X.shape[3], 1)
# CNN.ConvLSTM(X, y, epochs=epochs, name='ConvLSTM', num_GPU=num_GPU,
# optimizer='adam', batch_size=64, loss='categorical_crossentropy',
# metrics=['accuracy'], test_split_size=0.1, verbose=1)
# LSTM
# X = np.load('EEG.npy')
# print(X.shape)
# CNN.LSTMNN(X, y, epochs=10, name='LSTM', num_GPU=num_GPU,
# optimizer='adam', batch_size=32, loss='categorical_crossentropy',
# metrics=['accuracy'], test_split_size=0.3, verbose=1)
# LSTM fft
X = np.abs(np.load('stft-1D-100.npy'))
# print(X.shape)
# CNN.LSTMNN(X, y, epochs=1, name='LSTM', num_GPU=num_GPU,
# optimizer='adam', batch_size=32, loss='categorical_crossentropy',
# metrics=['accuracy'], test_split_size=0.3, verbose=1)
# print("Done!!!!!!!!!!!!!!!!!")
# channels all
X = np.load('EEG.npy')
####
####
X = reshape_1D_conv(X)
CNN.CNN1D(X, y, epochs=50, name='TimeDomain-allch', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
'''
# channels 7:21
#X = np.load('EEG.npy')
X = reshape_1D_conv(X[:, :, 7:21])
CNN.CNN1D(X, y, epochs=epochs, name='TimeDomain-ch0:22', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
print("Short Time Fourier 1D")
# Short Time Fourier 1D
X = np.abs(np.load('stft-1D-50.npy'))
X = reshape_1D_conv(X)
model = CNN.CNN1D(X, y, epochs=epochs, name='stft-1D-50', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
X = np.abs(np.load('stft-1D-100.npy'))
X = reshape_1D_conv(X)
model = CNN.CNN1D(X, y, epochs=epochs, name='stft-1D-100', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
X = np.abs(np.load('stft-1D-150.npy'))
X = reshape_1D_conv(X)
model = CNN.CNN1D(X, y, epochs=epochs, name='stft-1D-150', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
# Short Time Fourier 1D with log
print("Short Time Fourier 1D with log")
X = np.abs(np.load('stft-1D-50.npy'))
X = np.log10(X.clip(min=000000.1))
X = reshape_1D_conv(X)
model = CNN.CNN1D(X, y, epochs=epochs, name='stft-1D-50-log', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
X = np.abs(np.load('stft-1D-100.npy'))
X = np.log10(X.clip(min=000000.1))
X = reshape_1D_conv(X)
model = CNN.CNN1D(X, y, epochs=epochs, name='stft-1D-100-log', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
X = np.abs(np.load('stft-1D-150.npy'))
X = np.log10(X.clip(min=000000.1))
X = reshape_1D_conv(X)
model = CNN.CNN1D(X, y, epochs=epochs, name='stft-1D-150-log', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
'''
'''
# Short Time Fourier 2D
print("Short Time Fourier 2D")
X = np.abs(np.load('stft-2D-50.npy'))
model = CNN.CNN2D(X, y, epochs=epochs, name='stft-2D-50', num_GPU=num_GPU,
optimizer='rmsprop', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose,
shuffle = True)
X = np.abs(np.load('stft-2D-100.npy'))
model = CNN.CNN2D(X, y, epochs=epochs, name='stft-2D-100', num_GPU=num_GPU,
optimizer='rmsprop', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
X = np.abs(np.load('stft-2D-150.npy'))
model = CNN.CNN2D(X, y, epochs=epochs, name='stft-2D-150', num_GPU=num_GPU,
optimizer='rmsprop', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
'''
'''
# STFT 2D with Log Transformation
X = np.abs(np.load('stft-2D-50.npy'))
X = np.log10(X.clip(min=000000.1))
model = CNN.CNN2D(X, y, epochs=epochs, name='stft-2D-50-log', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
X = np.abs(np.load('stft-2D-100.npy'))
X = np.log10(X.clip(min=000000.1))
model = CNN.CNN2D(X, y, epochs=epochs, name='stft-2D-100-log', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
X = np.abs(np.load('stft-2D-150.npy'))
X = np.log10(X.clip(min=000000.1))
model = CNN.CNN2D(X, y, epochs=epochs, name='stft-2D-150-log', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
# Multitaper 1D
print("Multitaper 1D")
X = np.abs(np.load('mt-1D-50.npy'))
X = reshape_1D_conv(X)
model = CNN.CNN1D(X, y, epochs=epochs, name='mt-1D-50', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
X = np.abs(np.load('mt-1D-100.npy'))
X = reshape_1D_conv(X)
model = CNN.CNN1D(X, y, epochs=epochs, name='mt-1D-100', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
X = np.abs(np.load('mt-1D-150.npy'))
X = reshape_1D_conv(X)
model = CNN.CNN1D(X, y, epochs=epochs, name='mt-1D-150', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
# MultiTaper 1D with log
X = np.abs(np.load('mt-1D-50.npy'))
X = np.log10(X.clip(min=000000.1))
X = reshape_1D_conv(X)
model = CNN.CNN1D(X, y, epochs=epochs, name='mt-1D-50-log', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
X = np.abs(np.load('mt-1D-100.npy'))
X = np.log10(X.clip(min=000000.1))
X = reshape_1D_conv(X)
model = CNN.CNN1D(X, y, epochs=epochs, name='mt-1D-100-log', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
X = np.abs(np.load('mt-1D-150.npy'))
X = np.log10(X.clip(min=000000.1))
X = reshape_1D_conv(X)
model = CNN.CNN1D(X, y, epochs=epochs, name='mt-1D-150-log', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
# Multitaper 2D
X = np.abs(np.load('mt-2D-50.npy'))
model = CNN.CNN2D(X, y, epochs=epochs, name='mt-2D-50', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
X = np.abs(np.load('mt-2D-100.npy'))
model = CNN.CNN2D(X, y, epochs=epochs, name='mt-1D-100', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
X = np.abs(np.load('mt-2D-150.npy'))
model = CNN.CNN2D(X, y, epochs=epochs, name='mt-1D-150', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
# Multitaper 2D with Log
X = np.abs(np.load('stft-2D-50.npy'))
X = np.log10(X.clip(min=000000.1))
model = CNN.CNN2D(X, y, epochs=epochs, name='stft-2D-50-log', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
X = np.abs(np.load('stft-2D-100.npy'))
X = np.log10(X.clip(min=000000.1))
model = CNN.CNN2D(X, y, epochs=epochs, name='stft-2D-100-log', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
X = np.abs(np.load('stft-2D-150.npy'))
X = np.log10(X.clip(min=000000.1))
model = CNN.CNN2D(X, y, epochs=epochs, name='stft-2D-150-log', num_GPU=num_GPU,
optimizer='adam', batch_size=32, loss='categorical_crossentropy',
metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
# wavelet
# X = np.abs(np.load('cwt-1D-5.npy'))
# X = np.log10(X.clip(min=000000.1))
# X = reshape_1D_conv(X)
# model = CNN.CNN1D(X, y, epochs=epochs, name='cwt-1D-5-log', num_GPU=num_GPU,
# optimizer='adam', batch_size=32, loss='categorical_crossentropy',
# metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
# X = np.abs(np.load('cwt-2D-5.npy'))
# model = CNN.CNN2D(X, y, epochs=epochs, name='cwt-2D-5', num_GPU=num_GPU,
# optimizer='adam', batch_size=32, loss='categorical_crossentropy',
# metrics=['accuracy'], test_split_size=0.1, verbose=verbose)
'''