-
Notifications
You must be signed in to change notification settings - Fork 0
/
Attractor_Dehyadegari_GeneralStructure_PartitionedData_oneLoop.py
339 lines (309 loc) · 10.8 KB
/
Attractor_Dehyadegari_GeneralStructure_PartitionedData_oneLoop.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
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
# -*- coding: utf-8 -*-
"""
Created on Tue Dec 15 14:36:31 2020
@author: user
"""
# -*- coding: utf-8 -*-
"""
Created on Sun Dec 6 23:45:26 2020
@author: Shaghayegh Reza
"""
#------------------------------------------------------------------------------
from keras.models import load_model
from keras.layers import Activation as activation
from keras.layers import Concatenate
from keras.layers.core import Dense, Flatten , Dropout
from keras.layers.convolutional import Convolution2D
from keras.layers.normalization import BatchNormalization
from keras.layers.pooling import MaxPooling2D
from keras.models import Model
from keras import layers
from keras import Input
import numpy as np
from keras.utils.np_utils import to_categorical
from keras.optimizers import *
from keras import callbacks
from keras.callbacks import TensorBoard, CSVLogger
import os
import scipy.io
import h5py
import mat73
from keras import optimizers
from keras.layers import Lambda
import keras.backend as K
from tensorflow import keras
#------------------------------------------------------------------------------
def ReLU(x):
return x * (x > 0)
#------------------------------------------------------------------------------
#------------------------------------------------------------------------------
#input
LenPart=np.load('LenPart.npy')
NumParts=np.load('NumParts.npy')
i=0;
NamePartX='TrainPart'+ str(i) + '.npy'
NamePartY='LabelPart'+ str(i) + '.npy'
x_train=np.load(NamePartX)
y_train=np.load(NamePartY)
#------------------------------------------------------------------------------
#------------------------------------------------------------------------------
#points:will be diverged with relu function
# decrease lr? خیلی در عدم واگرایی موثر بود
NumH1=1000
NumH2=1000
NumH3=1000
NumH4=1000
NumH5=103
NumH6=100
NumH7=100
landa=0.7
NumData=x_train.shape[0]
NumOut=y_train.shape[1]
#-------------------------------------------------
#inputs:
L=len(x_train[0])
input_1 = Input(shape=(L,))
input_y2 = Input(shape=(NumH2,))
for i in range(NumParts):
y2before=0.1*np.random.rand(np.shape(x_train)[0],NumH2)
Name='y2beforePart'+ str(i)
np.save(Name,y2before)
del(y2before)
'''
input_y3 = Input(shape=(NumH3,))
for i in range(NumParts):
y3before=0.1*np.random.rand(np.shape(x_train)[0],NumH3)
Name='y3beforePart'+ str(i)
np.save(Name,y3before)
del(y3before)
y1before=0.1*np.random.rand(np.shape(x_train)[0],NumH1)
input_y1 = Input(shape=(NumH1,))
y4before=0.1*np.random.rand(np.shape(x_train)[0],NumH4)
input_y4 = Input(shape=(NumH4,))
y5before=0.1*np.random.rand(np.shape(x_train)[0],NumH5)
input_y5 = Input(shape=(NumH5,))
y6before=0.1*np.random.rand(np.shape(x_train)[0],NumH6)
input_y6 = Input(shape=(NumH6,))
y7before=0.1*np.random.rand(np.shape(x_train)[0],NumH7)
input_y7 = Input(shape=(NumH7,))
'''
#-------------------------------------------------
#layers:
def Merger(ip):
landa=0.7
a = ip[0]
b=ip[1]
return tf.keras.backend.tanh((1-landa)*a+ b)
#
v1=Dense(NumH1, activation="linear",name='v1')
v1a=Dense(NumH1, activation="tanh",name='v1a')
f1=Dense(NumH1, activation="linear",name='f1')
L1=Lambda(Merger,name='L1')
#
v2=Dense(NumH2, activation="linear",name='v2')
v2a=Dense(NumH2, activation="tanh",name='v2a')
f2=Dense(NumH2, activation="linear",name='f2')
L2=Lambda(Merger,name='L2')
#
v3=Dense(NumH3, activation="linear",name='v3')
v3a=Dense(NumH3, activation="tanh",name='v3a')
f3=Dense(NumH3, activation="linear",name='f3')
L3=Lambda(Merger,name='L3')
#
v4=Dense(NumH4, activation="linear",name='v4')
v4a=Dense(NumH4, activation="tanh",name='v4a')
f4=Dense(NumH4, activation="linear",name='f4')
L4=Lambda(Merger,name='L4')
#
v5=Dense(NumH5, activation="linear",name='v5')
v5a=Dense(NumH5, activation="tanh",name='v5a')
f5=Dense(NumH5, activation="linear",name='f5')
L5=Lambda(Merger,name='L5')
#
v6=Dense(NumH6, activation="linear",name='v6')
v6a=Dense(NumH6, activation="tanh",name='v6a')
f6=Dense(NumH6, activation="linear",name='f6')
L6=Lambda(Merger,name='L6')
#
v7=Dense(NumH7, activation="linear",name='v7')
v7a=Dense(NumH7, activation="tanh",name='v7a')
f7=Dense(NumH7, activation="linear",name='f7')
L7=Lambda(Merger,name='L7')
#-------------------------------------------------
#net:
x=input_1
y1=v1a(x)
#
y2_forward=v2(y1)
y2_before=input_y2
y2_feedback=f2(y2_before)
y2=L2([y2_forward,y2_feedback])
#
y3=v3a(y2)
#
y4=v4a(y3)
y5=v5a(y4)
#
z=y5
#
model=Model([input_1,input_y2], [y2_forward,z])
model.summary()
#-----------------------------------------------------------------------------
# train:
opt1 = optimizers.Adam(lr=0.00002, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0)
model.compile(loss={'v2': 'mean_squared_error','v5a': 'mean_squared_error'},
loss_weights={'v2': 0.2, 'v5a': 0.6},optimizer=opt1)
reduce_LR = callbacks.ReduceLROnPlateau(monitor='mean_squared_error', factor=0.1, patience=5, verbose=1, mode='max', min_delta=0.0001, cooldown=0, min_lr=0)
#
Feedback2 = Model(inputs=model.input,outputs=model.get_layer('f2').output)
#Feedback3 = Model(inputs=model.input,outputs=model.get_layer('f3').output)
#-----------------------------------------------------------------------------
for i in range(NumParts):
print(i)
NamePartX='TrainPart'+ str(i) + '.npy'
NamePartY='LabelPart'+ str(i) + '.npy'
Name_y2before='y2beforePart'+ str(i) + '.npy'
Name_y2new='y2new'+ str(i) + '.npy'
x_train=np.load(NamePartX)
y_train=np.load(NamePartY)
y2before=np.load(Name_y2before)
y2forward=model.predict([x_train,y2before])[0]
y2before=np.tanh(y2forward)
y2feedback=Feedback2.predict([x_train,y2before,])
y2=np.tanh((1-landa)*y2forward+y2feedback)
y2feedback=Feedback2.predict([x_train,y2])
y2new=np.tanh((1-landa)*y2forward+y2feedback)
y2before=y2
np.save(Name_y2before,y2before)
np.save(Name_y2new,y2new)
#-----------------
for j in range(100):
#print(j)
model.save('model.h5')
for i in range(NumParts):
print('i',j,i)
NamePartX='TrainPart'+ str(i) + '.npy'
NamePartY='LabelPart'+ str(i) + '.npy'
Name_y2before='y2beforePart'+ str(i) + '.npy'
Name_y2new='y2new'+ str(i) + '.npy'
x_train=np.load(NamePartX)
y_train=np.load(NamePartY)
y2before=np.load(Name_y2before)
y2new=np.load(Name_y2new)
#
history = model.fit([x_train,y2before], [y2new,y_train],epochs=1,shuffle=True,callbacks= [reduce_LR])
#
y2forward=model.predict([x_train,y2before])[0]
y2feedback=Feedback2.predict([x_train,y2before])
y2=np.tanh((1-landa)*y2forward+y2feedback)
y2feedback=Feedback2.predict([x_train,y2])
y2new=np.tanh((1-landa)*y2forward+y2feedback)
#
y2before=y2
#
np.save(Name_y2before,y2before)
np.save(Name_y2new,y2new)
#-------------------------------------------------
#test on train data
y2before=0.1*np.random.rand(np.shape(x_train)[0],NumH2)
y3before=0.1*np.random.rand(np.shape(x_train)[0],NumH3)
Feedback2 = Model(inputs=model.input,outputs=model.get_layer('f2').output)
Feedback3 = Model(inputs=model.input,outputs=model.get_layer('f3').output)
#
y2forward=model.predict([x_train,y2before,y3before])[0]
y2before=np.tanh(y2forward)
y2feedback=Feedback2.predict([x_train,y2before,y3before])
y2=np.tanh((1-landa)*y2forward+y2feedback)
#
y3forward=model.predict([x_train,y2before,y3before])[1]
y3before=np.tanh(y3forward)
y3feedback=Feedback3.predict([x_train,y2before,y3before])
y3=np.tanh((1-landa)*y3forward+y3feedback)
#
y2before=y2
y3before=y3
#
m1=model.predict([x_train,y2before,y3before])[2]
for i in range(5):
#
y2forward=model.predict([x_train,y2before,y3before])[0]
y2feedback=Feedback2.predict([x_train,y2before,y3before])
y2=np.tanh((1-landa)*y2forward+y2feedback)
#
y3forward=model.predict([x_train,y2before,y3before])[1]
y3feedback=Feedback3.predict([x_train,y2before,y3before])
y3=np.tanh((1-landa)*y3forward+y3feedback)
#
y2before=y2
y3before=y3
#
m5=model.predict([x_train,y2before,y3before])[2]
mse1= ((m1 - y_train) ** 2).mean(axis=None)
mse5= ((m5 - y_train) ** 2).mean(axis=None)
#classification error
A1=np.zeros(22216)
A5=np.zeros(22216)
for i in range(22216):
A1[i]=np.argmax(m1[i,:])
A5[i]=np.argmax(m5[i,:])
n1=sum(1 for i, j in zip(A1,labels) if i != j) /22216*100
n5=sum(1 for i, j in zip(A5,labels) if i != j) /22216*100
#------------------------------------------------------------------------------
import tensorflow as tf
import keras
from keras.models import load_model
model = load_model('model.h5', custom_objects={'tf': tf})
path='D:\Shapar\ShaghayeghUni\AfterPropozal\MyPrograms\EventExtraction\Keras'
PathTest=path+'\TestMat'
PathOut=path+'\TestOut'
import glob
D=glob.glob(PathTest+"/*.mat")
NumTest=len(D)
for i in range(NumTest):
CB= scipy.io.loadmat(D[i])
CB_value=CB['CB_context']
x_train=CB_value
#---------------------------------------
#test it
y2before=0.1*np.random.rand(np.shape(x_train)[0],NumH2)
y3before=0.1*np.random.rand(np.shape(x_train)[0],NumH3)
Feedback2 = Model(inputs=model.input,outputs=model.get_layer('f2').output)
Feedback3 = Model(inputs=model.input,outputs=model.get_layer('f3').output)
#
x_train=CB_value
y2forward=model.predict([x_train,y2before,y3before])[0]
y2before=np.tanh(y2forward)
y2feedback=Feedback2.predict([x_train,y2before,y3before])
y2=np.tanh((1-landa)*y2forward+y2feedback)
#
y3forward=model.predict([x_train,y2before,y3before])[1]
y3before=np.tanh(y3forward)
y3feedback=Feedback3.predict([x_train,y2before,y3before])
y3=np.tanh((1-landa)*y3forward+y3feedback)
#
y2before=y2
y3before=y3
#
m1=model.predict([x_train,y2before,y3before])[2]
'''
for j in range(5):
#
y2forward=model.predict([x_train,y2before,y3before])[0]
y2feedback=Feedback2.predict([x_train,y2before,y3before])
y2=np.tanh((1-landa)*y2forward+y2feedback)
#
y3forward=model.predict([x_train,y2before,y3before])[1]
y3feedback=Feedback3.predict([x_train,y2before,y3before])
y3=np.tanh((1-landa)*y3forward+y3feedback)
#
y2before=y2
y3before=y3
#
m5=model.predict([x_train,y2before,y3before])[2]
'''
y_predict=m1
#y_predict=m5
#---------------------------------------
D2=D[i].replace(PathTest, PathOut)
scipy.io.savemat(D2, {'lbl': y_predict})
#-----------------------------------------------------------------------------