-
Notifications
You must be signed in to change notification settings - Fork 11
/
utils.py
184 lines (155 loc) · 5.28 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
import numpy as np
import torch
import torch.nn as nn
import torch.nn.functional as F
from sklearn.metrics import accuracy_score
import torchvision.models as models
from collections import OrderedDict
import math
def feature_extractor(output_channel):
resnet18 = models.resnet18(pretrained=True)
#models.resnet18(pretrained=True) # pre-trained model under ImageNet
resnet18.avgpool = nn.AvgPool2d(3, 1) #for input size is 72*72
#resnet18.avgpool = nn.AvgPool2d(1, 1) #for input size is 32*32
num_ftrs = resnet18.fc.in_features
resnet18.fc = nn.Linear(num_ftrs, output_channel)
for param in resnet18.parameters():
param.requires_grad = True
return resnet18.cuda()
def l1_penalty(var):
return torch.abs(var)
def fix_nn(model, theta):
def k_param_fn(tmp_model, name=None):
if len(tmp_model._modules)!=0:
for(k,v) in tmp_model._modules.items():
if name is None:
k_param_fn(v, name=str(k))
else:
k_param_fn(v, name=str(name+'.'+k))
else:
for (k,v) in tmp_model._parameters.items():
if not isinstance(v,torch.Tensor):
continue
tmp_model._parameters[k] = theta[str(name + '.' + k)]
k_param_fn(model)
return model
class Hot_Plug(object):
def __init__(self, model):
self.model = model
self.params = OrderedDict(self.model.named_parameters())
def update(self, lr=0.1):
for param_name in self.params.keys():
path = param_name.split('.')
cursor = self.model
for module_name in path[:-1]:
cursor = cursor._modules[module_name]
if lr > 0:
cursor._parameters[path[-1]] = self.params[param_name] - lr*self.params[param_name].grad
else:
cursor._parameters[path[-1]] = self.params[param_name]
def restore(self):
self.update(lr=0)
class Critic_Network_MLP(nn.Module):
def __init__(self, h, hh):
super(Critic_Network_MLP, self).__init__()
self.fc1 = nn.Linear(h, hh)
self.fc2 = nn.Linear(hh, 1)
def forward(self, x):
x = torch.relu(self.fc1(x))
x = nn.functional.softplus(self.fc2(x))
return torch.mean(x)
class Critic_Network_Flatten_FTF(nn.Module):
def __init__(self, h, hh):
super(Critic_Network_Flatten_FTF, self).__init__()
self.fc1 = nn.Linear(h ** 2, hh)
self.fc2 = nn.Linear(hh, 1)
def forward(self, x):
x = torch.relu(self.fc1(x))
x = nn.functional.softplus(self.fc2(x))
return torch.mean(x)
def freeze_layer(model):
count = 0
para_optim = []
for k in model.children():
count +=1
# 6 should be changed properly
if count> 6:
for param in k.parameters():
para_optim.append(param)
else:
for param in k.parameters():
param.requires_grad = False
#print count
return para_optim
def classifier(class_num):
model = nn.Sequential(
nn.Linear(512, class_num),
)
def init_weights(m):
if type(m) == nn.Linear:
torch.nn.init.xavier_uniform(m.weight)
m.bias.data.fill_(0.01)
model.apply(init_weights)
return model.cuda()
def classifier_homo(class_num):
model = nn.Sequential(
nn.ReLU(),
nn.Linear(4096, class_num),
)
def init_weights(m):
if type(m) == nn.Linear:
torch.nn.init.xavier_uniform(m.weight)
m.bias.data.fill_(0.01)
model.apply(init_weights)
return model.cuda()
'''
def dg_net(x, param):
return torch.mean(F.softplus(F.linear(F.relu(F.linear(x,param[0],param[1])),param[2],param[3]))).cuda()
# x.view(1,-1) ---> add one or two FC layer to a scalar.
'''
def compute_accuracy(predictions, labels):
accuracy = accuracy_score(y_true=np.argmax(labels, axis=-1), y_pred=np.argmax(predictions, axis=-1))
return accuracy
def cos_dist(a,b):
eps = 1e-8
all_norm = a.norm()
signal = True
if signal:
a_norm = a / (a.norm(dim=1,keepdim=True)+eps)
b_norm = b / (b.norm(dim=1,keepdim=True)+eps)
else:
a_norm = a / all_norm
b_norm = b / all_norm
res = torch.mm(a_norm, b_norm.transpose(0,1))
return res
def write_log(log, log_path):
f = open(log_path, mode='a')
f.write(str(log))
f.write('\n')
f.close()
def unfold_label(labels, classes):
new_labels = []
assert len(np.unique(labels)) == classes
# minimum value of labels
mini = np.min(labels)
for index in range(len(labels)):
dump = np.full(shape=[classes], fill_value=0).astype(np.int8)
_class = int(labels[index]) - mini
dump[_class] = 1
new_labels.append(dump)
return np.array(new_labels)
def shuffle_data(samples, labels):
num = len(labels)
shuffle_index = np.random.permutation(np.arange(num))
shuffled_samples = samples[shuffle_index]
shuffled_labels = labels[shuffle_index]
return shuffled_samples, shuffled_labels
def learning_rate(init, epoch):
optim_factor = 0
if(epoch > 160):
optim_factor = 3
elif(epoch > 120):
optim_factor = 2
elif(epoch > 60):
optim_factor = 1
return init*math.pow(0.2, optim_factor)