-
Notifications
You must be signed in to change notification settings - Fork 0
/
trainnetwork_ann.py
executable file
·153 lines (120 loc) · 5.35 KB
/
trainnetwork_ann.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
import time as tiimmee
import keras
import numpy as np
import tensorflow as tf
from keras.layers import (Conv2D, Dense, Dropout, Flatten, Input, MaxPooling2D,
concatenate)
from keras.models import Model
from keras.utils import np_utils, plot_model
import loaddata
def importdata(dataset):
# Ｔhis function import the data from the database(csv file)
# And return the feature and the label of the problem
# 'dataset' is the name of the file
# 'train_feature, train_label, test_feature, test_label'
# is training data
# 'inputnum' is number of input data
# 'nb_classes' is the number of class
# load the data
questionpath = 'dataset/dataset-question1.txt'
answerpath = 'dataset/dataset-answer1.txt'
input_data, output_data = loaddata.load(questionpath, answerpath)
input_data = np.array(input_data)
output_data = np.array(output_data)
features = input_data[:, 1:]
labels = output_data
nb_classes = int(max(labels))+1
inputnum = features.shape[1]
# devide the input data, devide it to train data, test_data and left data
datashape = features.shape
num_train = int(datashape[0]*0.8)
num_test = int(datashape[0]*0.2)
train_feature = features[:num_train, :]
train_label = labels[:num_train]
test_feature = features[num_train:num_test+num_train, :]
test_label = labels[num_train:num_test+num_train]
# reshape the data
train_feature = train_feature.reshape(train_feature.shape[0], inputnum)
test_feature = test_feature.reshape(test_feature.shape[0], inputnum)
train_label = train_label.reshape(train_label.shape[0], 1)
test_label = test_label.reshape(test_label.shape[0], 1)
# translate the label data into onehot shape
train_label = np_utils.to_categorical(train_label, nb_classes)
test_label = np_utils.to_categorical(test_label, nb_classes)
return train_feature, train_label, test_feature, test_label, inputnum, nb_classes
def creatmodel_ann(inputnum, numOfJobs, layer=15):
# This function is used to create the ann model in keras
# 'inputnum' is number of input data
# 'numOfJobs' is the number of class
# Layer is the number of the hidden ann layers
# The output model is the created model used to train
# See http://keras-cn.readthedocs.io/en/latest/models/model/
sess = tf.InteractiveSession()
input1 = Input(shape=(inputnum, ), name='i1')
mid = Dense(200, activation='relu', use_bias=True)(input1)
for i in range(layer):
mid = Dense(150, activation='relu', use_bias=True)(mid)
mid = Dense(150, activation='relu', use_bias=True)(mid)
mid = Dense(150, activation='relu', use_bias=True)(mid)
mid = Dense(100, activation='relu', use_bias=True)(mid)
mid = Dense(80, activation='relu', use_bias=True)(mid)
mid = Dropout(0.25)(mid)
mid = Dense(50, activation='relu', use_bias=True)(mid)
mid = Dropout(0.25)(mid)
mid = Dense(20, activation='relu', use_bias=True)(mid)
out = Dense(numOfJobs, activation='softmax',
name='out', use_bias=True)(mid)
model = Model(input1, out)
sgd = keras.optimizers.SGD(
lr=0.11, momentum=0.0, decay=0.0, nesterov=False)
model.compile(
optimizer=sgd,
loss='mean_squared_error',
metrics=['accuracy'],
)
return model
def triannetwork(model, train_feature, train_label, test_feature, test_label,
batch_size, epochs):
# network training function
# see http://keras-cn.readthedocs.io/en/latest/models/model/
model.fit(
[train_feature], train_label, batch_size=batch_size, epochs=epochs,
verbose=1, validation_data=([test_feature], test_label)
)
return model
def savenetwork(model, name):
# This function is used to save the the medol trained above
time_now = int(tiimmee.time())
time_local = tiimmee.localtime(time_now)
time = tiimmee.strftime("%Y_%m_%d::%H_%M_%S", time_local)
print('current time is :', time)
savename = './model/' + 'ann_schedual_' + time + name+'.h5'
model.save(savename)
return savename
def main(dataset = 'featureandlable_traindata_m=8_n=8_timelow=6_timehight=30_numofloop=1000.csv'):
# ann parmater
layerofann = 15
# import the the data
# dataset = 'featureandlable_traindata_m=8_n=8_timelow=6_timehight=30_numofloop=1000.csv'
train_feature, train_label, test_feature, test_label, inputnum, nb_classes = importdata(
dataset)
print('successfully import the dataset :', dataset)
print('the shape of train_feature is :', train_feature.shape)
print('the shape of train_label is :', train_label.shape)
print('the shape of test_feature is :', test_feature.shape)
print('the shape of test_label is :', test_label.shape)
# create the nn model
model = creatmodel_ann(inputnum, nb_classes, layer=layerofann)
plot_model(model, to_file='model.png') # draw the figure of this model
# training parmater
batch_size = 49
epochs = 4000
# train the network
model = triannetwork(model, train_feature, train_label,
test_feature, test_label, batch_size, epochs)
# save the model
savename = "ann_layer"+str(layerofann) + '_' + dataset
savename = savenetwork(model, savename)
return test_feature, test_label, savename
if __name__ == '__main__':
test_feature, test_label, svaename = main()