-
Notifications
You must be signed in to change notification settings - Fork 0
/
neural.py
208 lines (179 loc) · 7.04 KB
/
neural.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
"""
The code containing neural network part, Skrlj 2019
"""
import torch
torch.manual_seed(123321)
import tqdm
import torch.nn as nn
from sklearn.preprocessing import OneHotEncoder
from torch.utils.data import DataLoader, Dataset
import logging
import numpy as np
np.random.seed(123321)
logging.basicConfig(format='%(asctime)s - %(message)s',
datefmt='%d-%b-%y %H:%M:%S')
logging.getLogger().setLevel(logging.INFO)
class E2EDatasetLoader(Dataset):
"""
A standard dataloader instance, note the csr subsetting.
"""
def __init__(self, features, targets=None, transform=None):
self.features = features.tocsr()
if not targets is None:
self.targets = targets.tocsr()
else:
self.targets = targets
def __len__(self):
return self.features.shape[0]
def __getitem__(self, index):
instance = torch.from_numpy(self.features[index, :].todense())
if self.targets is not None:
target = torch.from_numpy(self.targets[index].todense())
else:
target = None
return instance, target
def to_one_hot(lbx):
enc = OneHotEncoder(handle_unknown='ignore')
return enc.fit_transform(lbx.reshape(-1, 1))
class SimpleArch(nn.Module):
def __init__(self,
input_size,
dropout=0.1,
hidden_layer_size=10,
output_neurons=1):
"""
A simple architecture wrapper -- build with intuitive Sklearn-like API.
"""
super(SimpleArch, self).__init__()
self.h1 = nn.Linear(input_size, hidden_layer_size) ## prvi hidden
self.h2 = nn.Linear(
hidden_layer_size, hidden_layer_size
) ## drugi hidden -> dimenzija je ista kot od prvega.
self.h3 = nn.Linear(hidden_layer_size, 16) ## Tretji hidden ..
self.h4 = nn.Linear(16, output_neurons)
self.drop = nn.Dropout(dropout)
self.act = nn.ELU()
self.sigma = nn.Sigmoid()
def forward(self, x):
"""
The standard forward pass. See the original paper for the formal description of this part of the DRMs.
"""
out = self.h1(x)
out = self.drop(out)
out = self.act(out)
out = self.h2(out)
out = self.drop(out)
out = self.act(out)
out = self.h3(out)
out = self.drop(out)
out = self.act(out)
out = self.h4(out)
out = self.sigma(out)
return out
class E2EDNN:
"""
This is the main DRM class. The idea is to have a scikit-learn like interface for construction of ffNNs, capable of handling CSR-like inputs.
"""
def __init__(self,
batch_size=8,
num_epochs=10,
learning_rate=0.0001,
stopping_crit=10,
hidden_layer_size=30,
dropout=0.2,
file_type=None,
auto_depth=0):
self.device = torch.device(
'cuda' if torch.cuda.is_available() else 'cpu')
self.loss = torch.nn.BCELoss()
self.auto_depth = auto_depth
self.dropout = dropout
self.stopping_crit = stopping_crit
self.num_epochs = num_epochs
self.hidden_layer_size = hidden_layer_size
self.learning_rate = learning_rate
self.model = None
self.optimizer = None
self.num_params = None
def init_all(self, model, init_func, *params, **kwargs):
for p in model.parameters():
init_func(p, *params, **kwargs)
def fit(self, features, labels, onehot=False):
nun = len(np.unique(labels))
logging.info("Found {} unique labels.".format(nun))
labels = to_one_hot(labels)
train_dataset = E2EDatasetLoader(features, labels)
stopping_iteration = 0
loss = 1
current_loss = 0
self.model = SimpleArch(features.shape[1],
dropout=self.dropout,
hidden_layer_size=self.hidden_layer_size,
output_neurons=nun).to(self.device)
# self.init_all(self.model, torch.nn.init.normal_, mean=0., std=1)
# self.init_all(self.model, torch.nn.init.uniform_)
self.optimizer = torch.optim.Adam(self.model.parameters(),
lr=self.learning_rate)
self.num_params = sum(p.numel() for p in self.model.parameters())
logging.info("Number of parameters {}".format(self.num_params))
logging.info("Starting training for {} epochs".format(self.num_epochs))
for epoch in range(
self.num_epochs
): ## simple loss-based stopping works OK. TODO -> validation if more data is available.
if current_loss != loss:
current_loss = loss
else:
stopping_iteration += 1
if stopping_iteration > self.stopping_crit:
logging.info("Stopping reached!")
break
losses_per_batch = []
for i, (features, labels) in enumerate(train_dataset):
features = features.float().to(self.device)
labels = labels.float().to(self.device)
self.model.train()
outputs = self.model.forward(features).view(-1, 1)
labels = labels.view(-1, 1)
loss = self.loss(outputs, labels)
self.optimizer.zero_grad()
loss.backward()
self.optimizer.step()
losses_per_batch.append(float(loss))
mean_loss = np.mean(losses_per_batch)
logging.info("epoch {}, mean loss per batch {}".format(
epoch, mean_loss))
def predict(self, features, return_proba=False):
"""
Classic, sklearn-like predict method.
"""
test_dataset = E2EDatasetLoader(features, None)
predictions = []
self.model.eval()
with torch.no_grad():
for features, _ in test_dataset:
features = features.float().to(self.device)
representation = self.model.forward(features)
pred = representation.detach().cpu().numpy()[0]
predictions.append(pred)
if not return_proba:
a = [np.argmax(a_) for a_ in predictions] ## assumes 0 is 0
else:
a = []
for pred in predictions:
a.append(pred[1])
return np.array(a).flatten()
def predict_proba(self, features):
"""
It is also possible to obtain probabilistic outputs!
"""
test_dataset = E2EDatasetLoader(features, None)
predictions = []
self.model.eval()
with torch.no_grad():
for features, _ in test_dataset:
features = features.float().to(self.device)
representation = self.model.forward(features)
pred = representation.detach().cpu().numpy()[0]
predictions.append(pred)
a = [a_[1] for a_ in predictions]
return np.array(a).flatten()