-
Notifications
You must be signed in to change notification settings - Fork 2
/
utils.py
217 lines (179 loc) · 7.83 KB
/
utils.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
import numpy as np
import matplotlib.pyplot as plt
from tqdm import tqdm # Displays a progress bar
import torch
from torch import nn
from torch import optim
import torch.nn.functional as F
from torchvision import datasets, transforms
from torch.utils.data import Dataset, Subset, DataLoader, random_split, TensorDataset
from dataset import EnableDataset
import pickle
from sklearn.model_selection import KFold, StratifiedKFold,ShuffleSplit ,train_test_split
import copy
import os
import random
import pdb
import time
class trainclass():
def __init__(self,model,optimizer,databool,device,criterion,model_name,args=None):
self.model = model
self.optimizer = optimizer
self.data_bool = databool
self.device = device
self.criterion=criterion
self.model_name= model_name
self.args = args
def train(self, trainloader, num_epoch = 20, valloader=None, onehot=None): # Train the model
loss_history=[]
val_history=[]
correct = 0
print("Start training...")
self.model.train()
pre_loss=10000
for i in range(num_epoch):
running_loss = []
if self.args.mode_specific:
for batch, label, types, onehot in tqdm(trainloader,disable=self.data_bool):
onehot = onehot.to(self.device)
batch = batch.to(self.device)
label = label.to(self.device)
label = label -1 # indexing start from 1 (removing sitting conditon)
self.optimizer.zero_grad()
pred = self.model(batch,onehot)
loss = self.criterion(pred, label)
running_loss.append(loss.item())
loss.backward()
self.optimizer.step()
correct += (torch.argmax(pred,dim=1)==label).sum().item()
else:
for batch, label, types in tqdm(trainloader,disable=self.data_bool):
batch = batch.to(self.device)
label = label.to(self.device)
label = label -1 # indexing start from 1 (removing sitting conditon)
self.optimizer.zero_grad()
pred = self.model(batch)
loss = self.criterion(pred, label)
running_loss.append(loss.item())
loss.backward()
self.optimizer.step()
correct += (torch.argmax(pred,dim=1)==label).sum().item()
loss_mean= np.mean(running_loss)
loss_history.append(loss_mean)
if loss_mean< pre_loss:
pre_loss = loss_mean
torch.save(self.model.state_dict(), self.model_name)
print("*model saved*")
if self.args.val_on:
val_acc, _, _, _, _, _, _ = self.evaluate(valloader)
val_history.append(val_acc)
print("Epoch {} loss:{} val_acc:{}".format(i+1,np.mean(running_loss),val_acc))
else:
print("Epoch {} loss:{}".format(i+1,np.mean(running_loss)))
correct = 0
print("Done!")
return loss_history, val_history
def evaluate(self,loader):
self.model.eval()
correct = 0
steady_state_correct = 0
tot_steady_state = 0
transitional_correct = 0
tot_transitional = 0
preds=[]
tests=[]
class_correct = [0]*6
class_total = [0]*6
class_acc=[]
inferenceTime = 0.0
with torch.no_grad():
count = 0
totalloss = 0
if self.args.mode_specific:
for batch, label, types, onehot in tqdm(loader,disable=self.data_bool):
batch = batch.to(self.device)
label = label-1 # indexing start from 1 (removing sitting conditon)
label = label.to(self.device)
onehot = onehot.to(self.device)
if torch.cuda.is_available():
torch.cuda.synchronize()
beg = int(round(time.time()*1000))
pred = self.model(batch,onehot)
if torch.cuda.is_available():
torch.cuda.synchronize()
end = int(round(time.time()*1000))
inferenceTime += end - beg
totalloss += self.criterion(pred, label)
count +=1
preds.extend((torch.argmax(pred,dim=1)).tolist())
tests.extend(label.tolist())
correct += (torch.argmax(pred,dim=1)==label).sum().item()
steady_state_correct += (np.logical_and((torch.argmax(pred,dim=1) == label ).cpu(), types == 1)).sum().item()
tot_steady_state += (types == 1).sum().item()
transitional_correct += (np.logical_and((torch.argmax(pred,dim=1) == label ).cpu(), types == 0)).sum().item()
tot_transitional += (types == 0).sum().item()
for i in range(len(class_correct)):
class_correct[i] += (np.logical_and((torch.argmax(pred,dim=1) == label ).cpu(), label.cpu() == i)).sum().item()
class_total[i] += (label == i).sum().item()
else:
for batch, label, types in tqdm(loader,disable=self.data_bool):
batch = batch.to(self.device)
label = label-1 # indexing start from 1 (removing sitting conditon)
label = label.to(self.device)
if torch.cuda.is_available():
torch.cuda.synchronize()
beg = int(round(time.time()*1000))
pred = self.model(batch)
if torch.cuda.is_available():
torch.cuda.synchronize()
end = int(round(time.time()*1000))
inferenceTime += end - beg
totalloss += self.criterion(pred, label)
count +=1
preds.extend((torch.argmax(pred,dim=1)).tolist())
tests.extend(label.tolist())
correct += (torch.argmax(pred,dim=1)==label).sum().item()
steady_state_correct += (np.logical_and((torch.argmax(pred,dim=1) == label ).cpu(), types == 1)).sum().item()
tot_steady_state += (types == 1).sum().item()
transitional_correct += (np.logical_and((torch.argmax(pred,dim=1) == label ).cpu(), types == 0)).sum().item()
tot_transitional += (types == 0).sum().item()
for i in range(len(class_correct)):
class_correct[i] += (np.logical_and((torch.argmax(pred,dim=1) == label ).cpu(), label.cpu() == i)).sum().item()
class_total[i] += (label == i).sum().item()
acc = correct/len(loader.dataset)
for i in range(len(class_correct)):
if class_total[i] == 0:
print("Class {} has no samples".format(i))
else:
print("Class {} accuracy: {}".format(i, class_correct[i]/class_total[i]))
class_acc.append(class_correct[i]/class_total[i])
ss_acc = steady_state_correct/tot_steady_state if tot_steady_state != 0 else "No steady state samples used"
tr_acc = transitional_correct/tot_transitional if tot_transitional != 0 else "No transitional samples used"
print("Evaluation loss: {}".format(totalloss/count))
print("Evaluation accuracy: {}".format(acc))
print("Steady-state accuracy: {}".format(ss_acc))
print("Transistional accuracy: {}".format(tr_acc))
print("Inference Time: {} ms".format(inferenceTime / len(preds)))
return acc, ss_acc, tr_acc, preds, tests, class_acc, inferenceTime
def save_object(obj, filename):
with open(filename, 'wb') as output: # Overwrites any existing file.
pickle.dump(obj, output, pickle.HIGHEST_PROTOCOL)
def weight_classes(dataset):
trainloader = DataLoader(dataset, shuffle=False,batch_size=BATCH_SIZE)
classes = [0,0,0,0,0,0,0]
for data, labels in trainloader:
for x in range(labels.size()[0]):
classes[labels[x]] +=1
print(classes)
classes= classes[1:-1]
## with sample
weights=[]
sum_classes = np.sum(classes)
for idx in classes:
if idx != 0 :
weights.append(sum_classes/idx)
else:
continue
print(weights)
weights = torch.FloatTensor(weights)
return weights