/
SGAE_eval.py
149 lines (123 loc) · 5.25 KB
/
SGAE_eval.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
import time
import copy
import numpy as np
import mindspore as ms
from mindspore import nn
from mindspore import Model
from mindspore.train.callback import ModelCheckpoint, CheckpointConfig, LossMonitor, TimeMonitor, Callback
from sklearn.metrics import accuracy_score, average_precision_score, roc_auc_score
import pandas as pd
from tqdm import tqdm
from mindspore import save_checkpoint, load_checkpoint, load_param_into_net
from src.SGAE import SGAE
from dataloader import LoadDocumentData, LoadImageData, LoadTabularData
from mindspore import context, Tensor
from mindspore import dtype as mstype
from dataset import create_dataset
import os
import time
import argparse
import pandas as pd
from model_utils.config import config as cfg
def eval_tabular(params):
if params.ms_mode=="GRAPH":
context.set_context(mode=context.GRAPH_MODE)
else:
context.set_context(mode=context.PYNATIVE_MODE,pynative_synchronize=True)
context.set_context(device_target=params.device)
# Load data
x_train, y_train, x_val, y_val, x_test, y_test = LoadTabularData(params)
x_train_whole=Tensor(x_train,dtype=mstype.float32)
x_test_whole=Tensor(x_test,dtype=mstype.float32)
print(x_train_whole.shape)
global data_name
data_name=params.data_name
run_index=params.run_idx
# Start Train
# eval
model = SGAE(x_train_whole.shape[1], params.hidden_dim)
model.set_train(False)
param_dict = load_checkpoint(f"./saved_model/{params.data_name}_best_auc_pr_runs{run_index}.ckpt")
load_param_into_net(model, param_dict)
scores, _, _ = model(x_test_whole)
scores = scores.asnumpy()
auc = roc_auc_score(y_test, scores)
ap = average_precision_score(y_test, scores)
print(f'val finished, AUC={auc:.3f}, AP={ap:.3f}')
return {'AUC': f'{auc:.3f}', 'AP': f'{ap:.3f}'}
def eval_image(params):
context.set_context(mode=context.PYNATIVE_MODE,pynative_synchronize=True)
context.set_context(device_target="CPU")
# Load data
x_train, x_test, y_train, y_test = LoadImageData(params)
x_train_whole=Tensor(x_train,dtype=mstype.float32)
x_test_whole=Tensor(x_test,dtype=mstype.float32)
print(x_train_whole.shape)
global data_name
data_name=params.data_name
run_index=params.run_idx
# Start Train
# eval
model = SGAE(x_train_whole.shape[1], params.hidden_dim)
model.set_train(False)
param_dict = load_checkpoint(f"./saved_model/{params.data_name}_best_auc_pr_runs{run_index}.ckpt")
load_param_into_net(model, param_dict)
scores, _, _ = model(x_test_whole)
scores = scores.asnumpy()
auc = roc_auc_score(y_test, scores)
ap = average_precision_score(y_test, scores)
print(f'eval finished, AUC={auc:.3f}, AP={ap:.3f}')
return {'AUC': f'{auc:.3f}', 'AP': f'{ap:.3f}'}
def eval_document(params):
#global params_
#params_=params
context.set_context(mode=context.PYNATIVE_MODE,pynative_synchronize=True)
context.set_context(device_target="CPU")
# Load data
dataloader = LoadDocumentData(params)
# Experiment settings
auc = np.zeros((dataloader.class_num,))
ap = np.zeros((dataloader.class_num,))
#global data_name
data_name=params.data_name
run_idx=params.run_idx
for normal_idx in range(dataloader.class_num):
x_train, x_test, y_train, y_test = dataloader.preprocess(normal_idx)
x_train_whole=Tensor(x_train,dtype=mstype.float32)
x_test_whole=Tensor(x_test,dtype=mstype.float32)
#print(x_train_whole.shape)
model = SGAE(x_train_whole.shape[1], params.hidden_dim)
#model.set_train(False)
if params.verbose and normal_idx == 0 and run_idx == 0:
print(model)
param_dict = load_checkpoint(f"./saved_model/{params.data_name}_best_auc_pr_runs{run_idx}_normal{normal_idx}.ckpt")
load_param_into_net(model, param_dict)
scores, _, _ = model(x_test_whole)
scores = scores.asnumpy()
auc[normal_idx] = roc_auc_score(y_test, scores)
ap[normal_idx] = average_precision_score(y_test, scores)
print(f'this idx finished, AUC={np.mean(auc[normal_idx]):.3f}, AP={np.mean(ap[normal_idx]):.3f}')
return {'AUC': f'{np.mean(auc):.3f}', 'AP': f'{np.mean(ap):.3f}'}
if __name__ == '__main__':
start_time = time.time()
time_name = str(time.strftime("%m%d")) + '_' + str(time.time()).split(".")[1][-3:]
print(f'Time name is {time_name}')
print(os.getcwd())
# Total metrics
metrics = pd.DataFrame()
# Conduct one experiements
#args = parameter()
args=cfg
#print(f'Device is {args.device.type}-{args.cuda}')
if args.data_name in ['attack', 'bcsc', 'creditcard', 'diabetic', 'donor', 'intrusion', 'market']:
an_metrics_dict = eval_tabular(args)
elif args.data_name in ['reuters', '20news']:
an_metrics_dict = eval_document(args)
elif args.data_name in ['mnist']:
an_metrics_dict = eval_image(args)
metrics = pd.DataFrame(an_metrics_dict, index=[0])
metrics.to_csv(f'{args.out_dir}{args.model_name}_{args.data_name}_{time_name}.csv')
print(f'Finished!\nTotal time is {time.time()-start_time:.2f}s')
print(f'Current time is {time.strftime("%m%d_%H%M")}')
print(f'Results:')
print(metrics.sort_values('AUC', ascending=False))