-
Notifications
You must be signed in to change notification settings - Fork 0
/
NNutility.py
173 lines (130 loc) · 5.66 KB
/
NNutility.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
# -*- coding: utf-8 -*-
"""
Created on 22/01/21
@author: Cedric Beaulac
Clean place to drop NN Related stuff for LVS experiments
"""
import torch
import torch.nn.init as init
import torch.nn.functional as F
import torch.nn as nn
from torch.autograd import Variable
from torchvision import datasets, transforms
from torchvision.utils import save_image
import numpy as np
from numpy import *
import pandas as pd
from MEGA_Pytorch import *
####################################
# Defining VAEs
####################################
####################################
# PVAE: Probabilistic VAE
# Encoder has 1 hidden layer
# Decoder has one hidden layer
####################################
class PVAE(nn.Module):
def __init__(self, x_dim=10,h_dim=400,z_dim=20):
super(PVAE, self).__init__()
self.fc1 = nn.Linear(x_dim, h_dim)
self.fc21 = nn.Linear(h_dim, z_dim)
self.fc22 = nn.Linear(h_dim, z_dim)
self.fc3 = nn.Linear(z_dim, h_dim)
self.fc41 = nn.Linear(h_dim, x_dim)
self.fc42 = nn.Linear(h_dim, x_dim)
def forward(self, x):
mu, logvar = self.encode(x)
z = self.reparameterize(mu, logvar)
mux, logvarx = self.decode(z)
return mux,logvarx, mu, logvar
def encode(self, x):
h1 = F.relu(self.fc1(x))
return self.fc21(h1), self.fc22(h1)
def reparameterize(self, mu, logvar):
std = torch.exp(0.5*logvar)
eps = torch.randn_like(std)
return mu + eps*std
def decode(self, z):
h3 = F.relu(self.fc3(z))
return self.fc41(h3),self.fc42(h3)
####################################
# Define the training procedure
####################################
# Reconstruction + KL divergence losses summed over all elements and batch
def ploss_function(mux,logvarx, x, muz, logvarz,beta,device):
sigmax = logvarx.mul(0.5).exp_()
pxn = torch.distributions.normal.Normal(mux, sigmax)
logpx = torch.sum(pxn.log_prob(x.view(-1,784)))
KLD = -0.5 * torch.sum(1 + logvarz - muz.pow(2) - logvarz.exp())
return logpx - beta*KLD
def ptrain(args, model, device, data, optimizer, epoch,beta):
model.train()
perm = np.random.permutation(data.shape[0])
Alldata = torch.tensor(data[perm, :])
train_loss = 0
for i in range(0, args.Number_batch):
datas = Alldata[(i * args.batch_size):((i + 1) * args.batch_size),:].to(device)
optimizer.zero_grad()
mux,logvarx, muz, logvarz = model(datas)
loss = -ploss_function(mux,logvarx, datas, muz, logvarz,beta,device)
train_loss += loss.item()
loss.backward()
optimizer.step()
#train_loss /= data.shape[0]
#print('====> Epoch: {} Average loss: {:.4f}'.format(
# epoch, train_loss ))
def ptest(args, model, device, data, epoch,beta):
mux,logvarx, mu, logvar = model(data)
loss = -ploss_function(mux,logvarx, data, mu, logvar,beta,device)
n = min(data.size(0), 8)
comparison = torch.cat([data.view(-1, 1, 28, 28)[:n],
mux.view(-1, 1, 28, 28)[:n]])
save_image(comparison.cpu(),'results/reconstruction_' + str(epoch)
+ '.png', nrow=n)
print('====> Epoch: {} Average test loss: {:.4f}'.format(
epoch, loss/data.shape[0] ))
####################################
# Using MEGA for Regularization: Define the training procedure
####################################
# Reconstruction + KL divergence losses summed over all elements and batch
def ploss_function_regu(mux,logvarx, x, muz, logvarz,beta,alpha,device,model,ntz):
sigmax = logvarx.mul(0.5).exp_()
#sigmax = torch.ones(logvarx.shape)
#sigmax = sigmax.to(device)
pxn = torch.distributions.normal.Normal(mux, sigmax)
logpx = torch.sum(pxn.log_prob(x))
KLD = 0.5 * torch.sum(1 + logvarz - muz.pow(2) - logvarz.exp())
LDim = shape(muz)[1]
#VAE
NewPoint = np.random.normal(loc=np.zeros(LDim), scale=np.ones(LDim), size=(ntz, LDim))
NewPoint = torch.tensor(NewPoint)
NewPoint = NewPoint.type(torch.FloatTensor)
vaemux, vaelogvarx = model.decode(NewPoint)
Mega = MEGA(x,vaemux,torch.diag_embed(vaelogvarx))
return logpx + beta*KLD - alpha*(Mega[0]+Mega[1])
def ptrain_MEGA(args, model, device, data, optimizer, epoch,beta,alpha,ntz):
model.train()
perm = np.random.permutation(data.shape[0])
Alldata = torch.tensor(data[perm, :])
train_loss = 0
for i in range(0, args.Number_batch):
datas = Alldata[(i * args.batch_size):((i + 1) * args.batch_size),:].to(device)
optimizer.zero_grad()
mux,logvarx, muz, logvarz = model(datas)
loss = -ploss_function_regu(mux,logvarx, datas, muz, logvarz,beta,alpha,device,model,ntz)
train_loss += loss.item()
loss.backward()
optimizer.step()
#train_loss /= data.shape[0]
#print('====> Epoch: {} Average loss: {:.4f}'.format(
# epoch, train_loss ))
def ptest_MEGA(args, model, device, data, epoch,beta,alpha,ntz):
mux,logvarx, muz, logvarz = model(data)
loss = -ploss_function_regu(mux,logvarx, data, muz, logvarz,beta,alpha,device,model,ntz)
n = min(data.size(0), 8)
comparison = torch.cat([data.view(-1, 1, 28, 28)[:n],
mux.view(-1, 1, 28, 28)[:n]])
save_image(comparison.cpu(),'results/reconstruction_' + str(epoch)
+ '.png', nrow=n)
print('====> Epoch: {} Average test loss: {:.4f}'.format(
epoch, loss/data.shape[0] ))