-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathembedder.py
192 lines (135 loc) · 6.44 KB
/
embedder.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
import os
import copy
import torch
from src.utils import config2string
from src.transform import get_graph_drop_transform
from src.utils import compute_accuracy
from layers import GNN, Classifier
import os.path as osp
import statistics
class embedder:
def __init__(self, args):
self.args = args
self.device =torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')
torch.cuda.set_device(self.device)
self.config_str = config2string(args)
print("\n[Config] {}\n".format(self.config_str))
self.path = osp.join('/tmp', 'data', self.args.dataset)
self.transform1 = get_graph_drop_transform(drop_edge_p=args.de_1, drop_feat_p=args.df_1)
self.transform2 = get_graph_drop_transform(drop_edge_p=args.de_2, drop_feat_p=args.df_2)
if self.args.layers == 1:
self.hidden_layers = [self.args.dim]
elif self.args.layers == 2:
self.hidden_layers=[self.args.dim,self.args.dim]
elif self.args.layers == 3:
self.hidden_layers = [self.args.dim, self.args.dim,self.args.dim]
# For Evaluation
self.best_val = 0
self.epoch_list = [] # for epoch select
self.train_accs = [];
self.valid_accs = [];
self.test_accs = []
self.train_baccs = [];
self.valid_baccs = [];
self.test_baccs = []
self.train_f1 = [];
self.valid_f1 = [];
self.test_f1 = []
self.running_train_accs = [];
self.running_valid_accs = [];
self.running_test_accs = []
self.running_train_baccs = [];
self.running_valid_baccs = [];
self.running_test_baccs = []
self.running_train_f1 = [];
self.running_valid_f1 = [];
self.running_test_f1 = [];
def evaluate(self, batch_data, st):
# Classifier Accuracy
# Classifier Accuracy
self.model.eval()
_, preds = self.model.cls(batch_data)
train_acc, val_acc, test_acc,train_bacc,val_bacc,test_bacc,train_f1,val_f1,test_f1 = compute_accuracy(preds, batch_data.y, self.train_mask, self.val_mask,
self.test_mask)
self.running_train_accs.append(train_acc);
self.running_valid_accs.append(val_acc);
self.running_test_accs.append(test_acc)
self.running_train_baccs.append(train_bacc);
self.running_valid_baccs.append(val_bacc);
self.running_test_baccs.append(test_bacc)
self.running_train_f1.append(train_f1);
self.running_valid_f1.append(val_f1);
self.running_test_f1.append(test_f1)
if val_acc > self.best_val:
self.best_val = val_acc
self.cnt = 0
else:
self.cnt += 1
st += '| train_acc: {:.2f} | valid_acc : {:.2f} | test_acc : {:.2f}| ' \
'train_bacc: {:.2f} | valid_bacc : {:.2f} | test_bacc : {:.2f}' \
'| train_f1: {:.2f} | valid_f1 : {:.2f} | test_f1 : {:.2f}' \
.format(train_acc, val_acc, test_acc,train_bacc,val_bacc,test_bacc,train_f1,val_f1,test_f1)
print(st)
def save_results(self, fold):
train_acc, val_acc, test_acc= torch.tensor(self.running_train_accs), torch.tensor(
self.running_valid_accs), torch.tensor(self.running_test_accs)
train_bacc, val_bacc, test_bacc = torch.tensor(self.running_train_baccs), torch.tensor(
self.running_valid_baccs), torch.tensor(self.running_test_baccs)
train_f1, val_f1, test_f1 = torch.tensor(self.running_train_f1), torch.tensor(
self.running_valid_f1), torch.tensor(self.running_test_f1)
selected_epoch = val_acc.argmax()
best_train_acc = train_acc[selected_epoch]
best_val_acc = val_acc[selected_epoch]
best_test_acc = test_acc[selected_epoch]
best_train_bacc = train_bacc[selected_epoch]
best_val_bacc = val_bacc[selected_epoch]
best_test_bacc = test_bacc[selected_epoch]
best_train_f1 = train_f1[selected_epoch]
best_val_f1 = val_f1[selected_epoch]
best_test_f1 = test_f1[selected_epoch]
self.epoch_list.append(selected_epoch.item())
self.train_accs.append(best_train_acc.item());
self.valid_accs.append(best_val_acc.item());
self.test_accs.append(best_test_acc.item());
self.train_baccs.append(best_train_bacc.item());
self.valid_baccs.append(best_val_bacc.item());
self.test_baccs.append(best_test_bacc.item());
self.train_f1.append(best_train_f1.item());
self.valid_f1.append(best_val_f1.item());
self.test_f1.append(best_test_f1.item());
if fold + 1 != self.args.repetitions:
self.running_train_accs = [];
self.running_valid_accs = [];
self.running_test_accs = []
self.running_train_baccs = [];
self.running_valid_baccs = [];
self.running_test_baccs = []
self.running_train_f1 = [];
self.running_valid_f1 = [];
self.running_test_f1 = []
self.cnt = 0
self.best_val = 0
def summary(self):
if len(self.train_accs) == 1:
train_acc_mean = self.train_accs[0]
val_acc_mean = self.valid_accs[0]
test_acc_mean = self.test_accs[0]
val_f1_mean = self.valid_f1[0]
test_f1_mean = self.test_f1[0]
test_bacc_mean = self.test_baccs[0]
acc_CI = 0
bacc_CI = 0
f1_CI = 0
else:
train_acc_mean = statistics.mean(self.train_accs)
val_acc_mean = statistics.mean(self.valid_accs)
test_acc_mean = statistics.mean(self.test_accs)
val_f1_mean = statistics.mean(self.valid_f1)
test_f1_mean = statistics.mean(self.test_f1)
test_bacc_mean =statistics.mean(self.test_baccs)
acc_CI = (statistics.stdev(self.test_accs) / (self.args.repetitions ** (1 / 2)))
bacc_CI = (statistics.stdev(self.test_baccs) / (self.args.repetitions ** (1 / 2)))
f1_CI = (statistics.stdev(self.test_f1) / (self.args.repetitions ** (1 / 2)))
log= "** | test acc : {:.2f} +- {:.2f} | test bacc : {:.2f} +- {:.2f} | test f1 : {:.2f} +- {:.2f} |val acc: {:.2f} |val f1: {:.2f} |train acc: {:.2f} | **\n".format(
test_acc_mean, acc_CI, test_bacc_mean,bacc_CI,test_f1_mean,f1_CI,val_acc_mean,val_f1_mean, train_acc_mean)
print(log)