-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathlibs_comparison.py
460 lines (391 loc) · 22.4 KB
/
libs_comparison.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
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
# -------------------------------------
# Project: Learning to Compare: Relation Network for Few-Shot Learning
# Date: 2017.9.21
# Author: Flood Sung
# All Rights Reserved
# -------------------------------------
import gc
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.autograd import Variable
from torch.optim.lr_scheduler import StepLR
import numpy as np
import task_generator as tg
import os
import math
import argparse
import random
from src.efficient_kan import KAN
import CNNEncoder1
import vit
parser = argparse.ArgumentParser(description="One Shot Visual Recognition")
parser.add_argument("-f", "--feature_dim", type=int, default=128)
parser.add_argument("-r", "--relation_dim", type=int, default=8)
parser.add_argument("-w", "--class_num", type=int, default=2)
parser.add_argument("-s", "--sample_num_per_class", type=int, default=1)
parser.add_argument("-b", "--batch_num_per_class", type=int, default=4)
parser.add_argument("-e", "--episode", type=int, default=1000)
parser.add_argument("-t", "--test_episode", type=int, default=100)
parser.add_argument("-l", "--learning_rate", type=float, default=0.001)
parser.add_argument("-g", "--gpu", type=int, default=0)
parser.add_argument("-u", "--hidden_unit", type=int, default=10)
parser.add_argument("-c", "--cpu", type=int, default=0)
args = parser.parse_args()
# Hyper Parameters
FEATURE_DIM = args.feature_dim
RELATION_DIM = args.relation_dim
CLASS_NUM = args.class_num
SAMPLE_NUM_PER_CLASS = args.sample_num_per_class
BATCH_NUM_PER_CLASS = args.batch_num_per_class
EPISODE = args.episode
TEST_EPISODE = args.test_episode
LEARNING_RATE = args.learning_rate
GPU = args.gpu
HIDDEN_UNIT = args.hidden_unit
train_result = './train_result/'
label_list = ['Health', 'anomaly']
def weights_init(m):
classname = m.__class__.__name__
if classname.find('Conv') != -1:
n = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
m.weight.data.normal_(0, math.sqrt(2. / n))
if m.bias is not None:
m.bias.data.zero_()
elif classname.find('BatchNorm') != -1:
m.weight.data.fill_(1)
m.bias.data.zero_()
elif classname.find('Linear') != -1:
n = m.weight.size(1)
m.weight.data.normal_(0, 0.01)
m.bias.data = torch.ones(m.bias.data.size())
def main():
loos_result = []
loos_result_1 = []
accuray_result_1_1 = []
accuray_result_1_2 = []
accuray_result_1_3 = []
# Step 1: init data folders
print("init data folders")
# init character folders for dataset construction
# metatrain_character_folders,metatest_character_folders = tg.omniglot_character_folders(train,test,label_list)
# Step 2: init neural networks
print("init neural networks")
# 定义化网络
feature_encoder = CNNEncoder1.rsnet() # 特征提取
# relation_network = vit.ViT(image_size=28, patch_size=7, num_classes=2, dim=1024, depth=4, heads=8, mlp_dim=2048,
# dropout=0.1, emb_dropout=0.1) # 定义关系网络
relation_network = KAN([28 * 28, 128, 8])
kan = KAN([8 * 512, 512, 32, 2])
# 关系网络和特征提取模块的加载话
# feature_encoder.apply(weights_init)
# relation_network.apply(weights_init)
feature_encoder.cuda(GPU)
relation_network.cuda(GPU)
kan.cuda(GPU)
# 优化器的定义
feature_encoder_optim = torch.optim.Adam(feature_encoder.parameters(), lr=LEARNING_RATE)
feature_encoder_scheduler = StepLR(feature_encoder_optim, step_size=100000, gamma=0.5)
relation_network_optim = torch.optim.Adam(relation_network.parameters(), lr=LEARNING_RATE)
relation_network_scheduler = StepLR(relation_network_optim, step_size=100000, gamma=0.5)
# Step 3: build graph
print("Training...")
last_accuracy = 0
for episode in range(EPISODE):
# print('episode', episode)
feature_encoder_scheduler.step(episode)
relation_network_scheduler.step(episode)
# init dataset
# sample_dataloader is to obtain previous samples for compare
# batch_dataloader is to batch samples for training
degrees = random.choice([0, 90, 180, 270])
##第一个监测点 轴箱leftaxlebox
metatrain_character_folders_1 = ['../CWT-1000/motor/train/health',
'../CWT-1000/motor/train/anomaly1']
task_1 = tg.OmniglotTask(metatrain_character_folders_1, CLASS_NUM, SAMPLE_NUM_PER_CLASS, BATCH_NUM_PER_CLASS)
sample_dataloader_1 = tg.get_data_loader(task_1, num_per_class=SAMPLE_NUM_PER_CLASS, split="train",
shuffle=False, rotation=degrees)
batch_dataloader_1 = tg.get_data_loader(task_1, num_per_class=BATCH_NUM_PER_CLASS, split="test", shuffle=True,
rotation=degrees)
# sample datas
sample_dataloader_1 = iter(sample_dataloader_1)
samples_1, sample_labels_1 = next(sample_dataloader_1)
# print('samples',samples.shape)
batch_dataloader_1 = iter(batch_dataloader_1)
batches_1, batch_labels_1 = next(batch_dataloader_1)
## 特征提取
sample_features_1 = feature_encoder(Variable(samples_1).cuda(GPU)) # 5x64*5*5
batch_features_1 = feature_encoder(Variable(batches_1).cuda(GPU)) # 20x64*5*5
#######################################################################################
## 特征拼接
# print(sample_features_1.shape) # 2,128,28,28
# print(sample_labels_1.shape) # 2
sample_features_ext_1 = sample_features_1.unsqueeze(0).repeat(BATCH_NUM_PER_CLASS * CLASS_NUM, 1, 1, 1, 1)
sample_labels_1 = sample_labels_1.repeat(BATCH_NUM_PER_CLASS)
sample_labels_1 = sample_labels_1.long()
batch_features_ext_1 = batch_features_1.unsqueeze(0).repeat(SAMPLE_NUM_PER_CLASS * CLASS_NUM, 1, 1, 1, 1)
batch_features_ext_1 = torch.transpose(batch_features_ext_1, 0, 1)
# print(batch_features_ext_1.shape) # 38,2,128,28,28
relation_pairs_1 = torch.cat((sample_features_ext_1, batch_features_ext_1), 2)
# print(relation_pairs_1.shape) #38,2,256,28,28
# relation_pairs_1 = relation_pairs_1.view(-1, FEATURE_DIM * 2, 28, 28)
relation_pairs_1 = relation_pairs_1.view(-1, FEATURE_DIM * 4, 28 * 28)
##计算关系分数 kan
# print(relation_pairs_1.shape)
relations_1 = relation_network(relation_pairs_1)
relations_1 = relations_1.view(-1, 8 * 512)
relations_1 = kan(relations_1)
relations_1 = relations_1.view(-1, CLASS_NUM)
# print(relations_1.shape)
# print(relations_1.shape)
########################################################################################
# ##计算关系分数 kan
# # print(relation_pairs_1.shape)
# relations_1 = relation_network(relation_pairs_1)
# # print(relations_1.shape)
# relations_1 = relations_1.view(38, 8 * 512)
# relations_1 = kan(relations_1)
# relations_1 = relations_1.view(-1, CLASS_NUM)
# print(relations_1.shape)
# print(relations_1.shape)
# ########################################################################################
# relations_2 = relation_network(relation_pairs_2)
# relations_2 = relations_2.view(-1, CLASS_NUM)
# ########################################################################################
# relations_3 = relation_network(relation_pairs_3)
# relations_3 = relations_3.view(-1, CLASS_NUM)
mse = nn.MSELoss().cuda(GPU)
# 计算LOSS
batch_labels_1 = batch_labels_1.long()
# print(batches_1.shape)
# print(batch_labels_1.view(-1,1))
one_hot_labels_1 = torch.zeros(BATCH_NUM_PER_CLASS * CLASS_NUM, CLASS_NUM).scatter_(1,
batch_labels_1.view(-1, 1),
1).cuda(GPU)
# print(one_hot_labels_1.shape)
# print(relations_1.shape)
loss_1 = mse(relations_1, one_hot_labels_1)
########################################################################################
feature_encoder.zero_grad()
relation_network.zero_grad()
loss = loss_1
loss.backward()
torch.nn.utils.clip_grad_norm(feature_encoder.parameters(), 0.5)
torch.nn.utils.clip_grad_norm(relation_network.parameters(), 0.5)
feature_encoder_optim.step()
relation_network_optim.step()
loos_result_1.append(loss_1)
loos_result.append(loss)
if (episode + 1) % 50 == 0:
print("episode:", episode + 1, "loss", loss.item())
loos_result.append(loss)
if (episode + 1) % 100 == 0:
# test
print("Testing...1-1,motor type13",end='')
total_rewards_1_1 = 0
for i in range(TEST_EPISODE):
degrees = random.choice([0, 90, 180, 270])
metatest_character_folders1 = ['../CWT-1000/motor/test/health',
'../CWT-1000/motor/test/anomaly/anomalyTYPE3']
# '../CWT-1000/motor/test/anomaly/anomalyTYPE13']
# '../CWT-1000/motor/test/anomaly/anomalyTYPE14']
metatrain_character_folders1 = ['../CWT-1000/motor/train/health',
'../CWT-1000/motor/train/anomaly1']
task = tg.OmniglotTask(metatest_character_folders1, CLASS_NUM, SAMPLE_NUM_PER_CLASS,
SAMPLE_NUM_PER_CLASS, )
task1 = tg.OmniglotTask(metatrain_character_folders1, CLASS_NUM, SAMPLE_NUM_PER_CLASS,
BATCH_NUM_PER_CLASS)
sample_dataloader = tg.get_data_loader(task1, num_per_class=SAMPLE_NUM_PER_CLASS, split="train",
shuffle=False, rotation=degrees)
test_dataloader = tg.get_data_loader(task, num_per_class=SAMPLE_NUM_PER_CLASS, split="test",
shuffle=True, rotation=degrees)
sample_dataloader = iter(sample_dataloader)
sample_images, sample_labels = next(sample_dataloader)
test_dataloader = iter(test_dataloader)
test_images, test_labels = next(test_dataloader)
sample_features = feature_encoder(Variable(sample_images).cuda(GPU)) # 5x64
test_features = feature_encoder(Variable(test_images).cuda(GPU)) # 20x64
sample_features_ext = sample_features.unsqueeze(0).repeat(SAMPLE_NUM_PER_CLASS * CLASS_NUM, 1, 1, 1, 1)
test_features_ext = test_features.unsqueeze(0).repeat(SAMPLE_NUM_PER_CLASS * CLASS_NUM, 1, 1, 1, 1)
test_features_ext = torch.transpose(test_features_ext, 0, 1)
# print(torch.cat((sample_features_ext, test_features_ext), 2).shape) #2,2,256,28,28
# relation_pairs = torch.cat((sample_features_ext, test_features_ext), 2).view(-1, FEATURE_DIM * 2, 28,
# 28)
#
relation_pairs = torch.cat((sample_features_ext, test_features_ext), 2).view(-1,
FEATURE_DIM * 4, 28 * 28)
##kan
# relations1 = relation_network(relation_pairs)
# # print(relations1.shape)
# relations1 = relations1.view(2, 8 * 512)
# relations1 = kan(relations1)
# relations = relations1.view(-1, CLASS_NUM)
#
# # relations1 = relation_network(relation_pairs)
# # relations = relations1.view(-1, CLASS_NUM)
# bb = Variable(torch.zeros(CLASS_NUM)).cuda(GPU)
# transformer
relations1 = relation_network(relation_pairs)
relations1 = relations1.view(2, 8 * 512)
relations1 = kan(relations1)
relations = relations1.view(-1, CLASS_NUM)
bb = Variable(torch.zeros(CLASS_NUM)).cuda(GPU)
# print(relations)
for j in range(len(relations)):
# print(relations[j][0])
if relations[j][0] > 0.9:
bb[j] = 0
else:
bb[j] = 1
# _,predict_labels = torch.max(relations.data,1)
predict_labels = bb.cpu()
rewards = [1 if predict_labels[j] == test_labels[j] else 0 for j in range(CLASS_NUM)]
total_rewards_1_1 += np.sum(rewards)
test_accuracy_1_1 = total_rewards_1_1 / 1.0 / CLASS_NUM / TEST_EPISODE
accuray_result_1_1.append(test_accuracy_1_1)
print(" test accuracy 1-1:", test_accuracy_1_1)
print("Testing...1-2,lefaxlebox type14", end='')
total_rewards_1_2 = 0
for i in range(TEST_EPISODE):
degrees = random.choice([0, 90, 180, 270])
metatest_character_folders1 = ['../CWT-1000/motor/test/health',
'../CWT-1000/motor/test/anomaly/anomalyTYPE2']
# '../CWT-1000/motor/test/anomaly/anomalyTYPE13']
# '../CWT-1000/motor/test/anomaly/anomalyTYPE14']
metatrain_character_folders1 = ['../CWT-1000/motor/train/health',
'../CWT-1000/motor/train/anomaly1']
task = tg.OmniglotTask(metatest_character_folders1, CLASS_NUM, SAMPLE_NUM_PER_CLASS,
SAMPLE_NUM_PER_CLASS, )
task1 = tg.OmniglotTask(metatrain_character_folders1, CLASS_NUM, SAMPLE_NUM_PER_CLASS,
BATCH_NUM_PER_CLASS)
sample_dataloader = tg.get_data_loader(task1, num_per_class=SAMPLE_NUM_PER_CLASS, split="train",
shuffle=False, rotation=degrees)
test_dataloader = tg.get_data_loader(task, num_per_class=SAMPLE_NUM_PER_CLASS, split="test",
shuffle=True, rotation=degrees)
sample_dataloader = iter(sample_dataloader)
sample_images, sample_labels = next(sample_dataloader)
test_dataloader = iter(test_dataloader)
test_images, test_labels = next(test_dataloader)
sample_features = feature_encoder(Variable(sample_images).cuda(GPU)) # 5x64
test_features = feature_encoder(Variable(test_images).cuda(GPU)) # 20x64
sample_features_ext = sample_features.unsqueeze(0).repeat(SAMPLE_NUM_PER_CLASS * CLASS_NUM, 1, 1, 1, 1)
test_features_ext = test_features.unsqueeze(0).repeat(SAMPLE_NUM_PER_CLASS * CLASS_NUM, 1, 1, 1, 1)
test_features_ext = torch.transpose(test_features_ext, 0, 1)
# print(torch.cat((sample_features_ext, test_features_ext), 2).shape) #2,2,256,28,28
# relation_pairs = torch.cat((sample_features_ext, test_features_ext), 2).view(-1, FEATURE_DIM * 2, 28,
# 28)
#
relation_pairs = torch.cat((sample_features_ext, test_features_ext), 2).view(-1,
FEATURE_DIM * 4, 28 * 28)
##kan
# relations1 = relation_network(relation_pairs)
# # print(relations1.shape)
# relations1 = relations1.view(2, 8 * 512)
# relations1 = kan(relations1)
# relations = relations1.view(-1, CLASS_NUM)
#
# # relations1 = relation_network(relation_pairs)
# # relations = relations1.view(-1, CLASS_NUM)
# bb = Variable(torch.zeros(CLASS_NUM)).cuda(GPU)
# transformer
relations1 = relation_network(relation_pairs)
relations1 = relations1.view(2, 8 * 512)
relations1 = kan(relations1)
relations = relations1.view(-1, CLASS_NUM)
bb = Variable(torch.zeros(CLASS_NUM)).cuda(GPU)
# print(relations)
for j in range(len(relations)):
if relations[j][0] > 0.9:
bb[j] = 0
else:
bb[j] = 1
# _,predict_labels = torch.max(relations.data,1)
predict_labels = bb.cpu()
rewards = [1 if predict_labels[j] == test_labels[j] else 0 for j in range(CLASS_NUM)]
total_rewards_1_2 += np.sum(rewards)
test_accuracy_1_2 = total_rewards_1_2 / 1.0 / CLASS_NUM / TEST_EPISODE
accuray_result_1_2.append(test_accuracy_1_2)
print(" test accuracy 1-2:", test_accuracy_1_2)
print("Testing...1-3,lefaxlebox type15", end='')
total_rewards_1_3 = 0
for i in range(TEST_EPISODE):
degrees = random.choice([0, 90, 180, 270])
metatest_character_folders1 = ['../CWT-1000/motor/test/health',
'../CWT-1000/motor/test/anomaly/anomalyTYPE1']
# '../CWT-1000/motor/test/anomaly/anomalyTYPE13']
# '../CWT-1000/motor/test/anomaly/anomalyTYPE14']
metatrain_character_folders1 = ['../CWT-1000/motor/train/health',
'../CWT-1000/motor/train/anomaly1']
task = tg.OmniglotTask(metatest_character_folders1, CLASS_NUM, SAMPLE_NUM_PER_CLASS,
SAMPLE_NUM_PER_CLASS, )
task1 = tg.OmniglotTask(metatrain_character_folders1, CLASS_NUM, SAMPLE_NUM_PER_CLASS,
BATCH_NUM_PER_CLASS)
sample_dataloader = tg.get_data_loader(task1, num_per_class=SAMPLE_NUM_PER_CLASS, split="train",
shuffle=False, rotation=degrees)
test_dataloader = tg.get_data_loader(task, num_per_class=SAMPLE_NUM_PER_CLASS, split="test",
shuffle=True, rotation=degrees)
sample_dataloader = iter(sample_dataloader)
sample_images, sample_labels = next(sample_dataloader)
test_dataloader = iter(test_dataloader)
test_images, test_labels = next(test_dataloader)
sample_features = feature_encoder(Variable(sample_images).cuda(GPU)) # 5x64
test_features = feature_encoder(Variable(test_images).cuda(GPU)) # 20x64
sample_features_ext = sample_features.unsqueeze(0).repeat(SAMPLE_NUM_PER_CLASS * CLASS_NUM, 1, 1, 1, 1)
test_features_ext = test_features.unsqueeze(0).repeat(SAMPLE_NUM_PER_CLASS * CLASS_NUM, 1, 1, 1, 1)
test_features_ext = torch.transpose(test_features_ext, 0, 1)
# print(torch.cat((sample_features_ext, test_features_ext), 2).shape) #2,2,256,28,28
# relation_pairs = torch.cat((sample_features_ext, test_features_ext), 2).view(-1, FEATURE_DIM * 2, 28,
# 28)
#
relation_pairs = torch.cat((sample_features_ext, test_features_ext), 2).view(-1,
FEATURE_DIM * 4, 28 * 28)
##kan
# relations1 = relation_network(relation_pairs)
# # print(relations1.shape)
# relations1 = relations1.view(2, 8 * 512)
# relations1 = kan(relations1)
# relations = relations1.view(-1, CLASS_NUM)
#
# # relations1 = relation_network(relation_pairs)
# # relations = relations1.view(-1, CLASS_NUM)
# bb = Variable(torch.zeros(CLASS_NUM)).cuda(GPU)
# transformer
relations1 = relation_network(relation_pairs)
relations1 = relations1.view(2, 8 * 512)
relations1 = kan(relations1)
relations = relations1.view(-1, CLASS_NUM)
bb = Variable(torch.zeros(CLASS_NUM)).cuda(GPU)
# print(relations)
for j in range(len(relations)):
if relations[j][0] > 0.9:
bb[j] = 0
else:
bb[j] = 1
# _,predict_labels = torch.max(relations.data,1)
predict_labels = bb.cpu()
rewards = [1 if predict_labels[j] == test_labels[j] else 0 for j in range(CLASS_NUM)]
total_rewards_1_3 += np.sum(rewards)
test_accuracy_1_3 = total_rewards_1_3 / 1.0 / CLASS_NUM / TEST_EPISODE
accuray_result_1_3.append(test_accuracy_1_3)
print(" test accuracy 1-3:", test_accuracy_1_3)
#############################################################################################################
###总精度
test_accuracy = test_accuracy_1_1 + test_accuracy_1_2 + test_accuracy_1_3
test_accuracy = test_accuracy / 3.0
print("TOtal_ accuracy:", test_accuracy)
if test_accuracy >= last_accuracy:
# save networks
torch.save(feature_encoder.state_dict(),
str("./models1/motor_feature_encoder_" + str(CLASS_NUM) + "way_" + str(
SAMPLE_NUM_PER_CLASS) + "shot.pkl"))
torch.save(relation_network.state_dict(),
str("./models1/motor_relation_network_" + str(CLASS_NUM) + "way_" + str(
SAMPLE_NUM_PER_CLASS) + "shot.pkl"))
torch.save(kan.state_dict(),
str("./models1/motor_relation_network_2" + str(CLASS_NUM) + "way_" + str(
SAMPLE_NUM_PER_CLASS) + "shot.pkl"))
print("save networks for episode:", episode)
last_accuracy = test_accuracy
return loos_result_1
if __name__ == '__main__':
loos_result_1 = main()