-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathmain.py
144 lines (119 loc) · 5.14 KB
/
main.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
# -*- coding: utf-8 -*-
import numpy as np
from funcCNN import *
from CG3Model import GCNModel
import tensorflow as tf
import time
from train import HGCN_Model
import sys
def GCNevaluate(mask1, labels1):
t_test = time.time()
outs_val = sess.run([GCNmodel.loss, GCNmodel.accuracy], feed_dict={labels: labels1, mask: mask1})
return outs_val[0], outs_val[1], (time.time() - t_test)
dataset_name = 'citeseer'
seed = 123
hidden_num = 1024
learning_rate = 0.01
epochs = 1000
dropout_all = 0.6
weight_decay = 0.1
adj, features, y_train, y_val, y_test, train_mask, val_mask, test_mask = LoadData(dataset_name)
#########################
num_classes = np.shape(y_train)[1]
num_inst = np.shape(y_train)[0]
features = preprocess_features(features)
feature_sp = tf.SparseTensor(features[0], np.array(features[1], dtype='float32'), features[2])
#########
input_dim = features[2][1]
support = preprocess_adj(adj)
num_inst = features[2][0]
trtemask = processmask(train_mask)
placeholders = {
'support': tf.sparse_placeholder(tf.float32),
'features': tf.sparse_placeholder(tf.float32, shape=tf.constant(features[2], dtype=tf.int64)),
'labels': tf.placeholder(tf.float32, shape=(None, y_train.shape[1])),
'labels_mask': tf.placeholder(tf.int32),
'dropout': tf.placeholder_with_default(0., shape=()),
'num_features_nonzero': tf.placeholder(tf.int32),
}
dp_fea0 = [placeholders['dropout'], placeholders['num_features_nonzero']]
mask = tf.placeholder("int32", [None])
labels = tf.placeholder("float", [None, num_classes])
paras = dict()
paras['hidden_num'] = hidden_num
paras['weight_decay'] = weight_decay
paras['dataset'] = dataset_name
HGCNModel = HGCN_Model(placeholders, paras)
y_dim1 = np.argmax(y_train, axis = 1)
y_dim = np.ones([num_inst]) * -1
tr_idx = np.argwhere(np.sum(y_train, axis = 1) > 0)[:, 0]
y_dim[tr_idx] = y_dim1[tr_idx]
intra_class_idx = []
for i in range(num_classes):
intra_class_idx.append(np.argwhere(y_dim == i)[:, 0])
train_mat01 = CalCLass01Mat(y_train, train_mask)
mats_intra_inter = CalIntraClassMat01(y_dim1[tr_idx])
num_labeled = int(np.sum(y_train))
mats_intra_inter[0] += np.eye(num_labeled)
np.random.seed(seed)
tf.set_random_seed(seed)
GCNmodel = GCNModel(feature_sp = feature_sp, learning_rate = learning_rate,
num_classes = num_classes, support = placeholders['support'],
h = hidden_num, input_dim = input_dim,
HGCN = HGCNModel, train_idx = tr_idx,
trtemask = trtemask, labels = labels, mask = mask,
dp_fea0 = dp_fea0, edge_pos = support[0], train_mat01 = train_mat01,
mat01_tr_te = mats_intra_inter, weight_decay = weight_decay)
sess=tf.Session()
sess.run(tf.global_variables_initializer())
test_accs = []
train_losses = []
train_accs = []
test_losses = []
val_accs = []
val_losses = []
# real_test_accs = []
# real_test_loss_acc = []
for epoch in range(epochs):
###train
feed_dict = construct_feed_dict_1(support, features, y_train, train_mask, placeholders, mask, labels)
feed_dict.update({placeholders['dropout']: dropout_all})
outs = sess.run([GCNmodel.opt_op, GCNmodel.loss, GCNmodel.accuracy], feed_dict=feed_dict)
###
if epoch % 1 == 0:
###test
feed_dict.update({mask: test_mask})
feed_dict.update({labels: y_test})
feed_dict.update({placeholders['dropout']: 0})
outs_val = sess.run([GCNmodel.loss, GCNmodel.accuracy], feed_dict=feed_dict)
################
#validation
feed_dict.update({mask: val_mask})
feed_dict.update({labels: y_val})
feed_dict.update({placeholders['dropout']: 0})
outs_validation = sess.run([GCNmodel.loss, GCNmodel.accuracy], feed_dict=feed_dict)
#########
print("Epoch:", '%04d' % (epoch + 1),
# "train_loss=", "{:.5f}".format(outs[1]),
"train_accuracy=", "{:.5f}".format(outs[2]),
"test_accuracy=", "{:.5f}".format(outs_val[1]),
"val_accuracy=", "{:.5f}".format(outs_validation[1]),
"test_loss=", "{:.5f}".format(outs_val[0]))
train_accs.append(outs[2])
# scio.savemat('train_accs.mat',{'train_accs':train_accs})
test_accs.append(outs_val[1])
# scio.savemat('test_accs.mat',{'test_accs':test_accs})
test_losses.append(outs_val[0])
# scio.savemat('test_losses.mat',{'test_losses':test_losses})
train_losses.append(outs[1])
# scio.savemat('train_losses.mat',{'train_losses':train_losses})
val_accs.append(outs_validation[1])
# scio.savemat('val_accs.mat',{'val_accs':val_accs})
val_losses.append(outs_validation[0])
# scio.savemat('val_losses.mat',{'val_losses':val_losses})
val_max = np.argmax(np.array(val_accs))
print(test_accs[val_max], np.max(test_accs))
print("test result:", test_accs[val_max])
#
# real_test_accs.append(test_accs[val_max])
# scio.savemat('real_test_accs.mat',{'real_test_accs':real_test_accs})