implement VAE

.gitignore

@@ -1,2 +1,3 @@
 mnist/data
+VAE/data
 *.pyc

VAE/README.md

@@ -0,0 +1,13 @@
+# Basic VAE Example
+
+This is an improved implementation of the paper [Stochastic Gradient VB and the
+Variational Auto-Encoder](http://arxiv.org/abs/1312.6114) by Kingma and Welling.
+It uses ReLUs and the adam optimizer, instead of sigmoids and adagrad. These changes make the network converge much faster.
+
+We reuse the data preparation script of the MNIST experiment
+
+```bash
+pip install -r requirements.txt
+python ../mnist/data.py
+python main.py
+```

VAE/main.py

@@ -0,0 +1,134 @@
+from __future__ import print_function
+import os
+import torch
+import torch.nn as nn
+import torch.optim as optim
+from torch.autograd import Variable
+
+cuda = torch.cuda.is_available()
+
+print('Running with CUDA: {0}'.format(cuda))
+
+
+def print_header(msg):
+    print('===>', msg)
+
+
+assert os.path.exists('data/processed/training.pt'), \
+    "Please run python ../mnist/data.py before starting the VAE."
+
+# Data
+print_header('Loading data')
+with open('data/processed/training.pt', 'rb') as f:
+    training_set = torch.load(f)
+with open('data/processed/test.pt', 'rb') as f:
+    test_set = torch.load(f)
+
+training_data = training_set[0].view(-1, 784).div(255)
+test_data = test_set[0].view(-1, 784).div(255)
+
+del training_set
+del test_set
+
+# Model
+print_header('Building model')
+
+
+class VAE(nn.Container):
+    def __init__(self):
+        super().__init__()
+
+        self.fc1 = nn.Linear(784, 400)
+        self.relu = nn.ReLU()
+        self.fc21 = nn.Linear(400, 20)
+        self.fc22 = nn.Linear(400, 20)
+        self.fc3 = nn.Linear(20, 400)
+        self.fc4 = nn.Linear(400, 784)
+        self.sigmoid = nn.Sigmoid()
+
+    def encode(self, x):
+        h1 = self.relu(self.fc1(x))
+        return self.fc21(h1), self.fc22(h1)
+
+    def reparametrize(self, mu, logvar):
+        std = logvar.mul(0.5).exp_()
+        eps = Variable(torch.randn(std.size()), requires_grad=False)
+        return eps.mul(std).add_(mu)
+
+    def decode(self, z):
+        h3 = self.relu(self.fc3(z))
+        return self.sigmoid(self.fc4(h3))
+
+    def forward(self, x):
+        mu, logvar = self.encode(x)
+        z = self.reparametrize(mu, logvar)
+        return self.decode(z), mu, logvar
+
+
+model = VAE()
+if cuda is True:
+    model.cuda()
+
+reconstruction_function = nn.BCELoss()
+reconstruction_function.size_average = False
+
+
+def loss_function(recon_x, x, mu, logvar):
+    BCE = reconstruction_function(recon_x, x)
+
+    # Appendix B from VAE paper: 0.5 * sum(1 + log(sigma^2) - mu^2 - sigma^2)
+    KLD_element = mu.pow(2).add_(logvar.exp()).mul_(-1).add_(1).add_(logvar)
+    KLD = torch.sum(KLD_element).mul_(-0.5)
+
+    return BCE + KLD
+
+
+# Training settings
+BATCH_SIZE = 150
+TEST_BATCH_SIZE = 1000
+NUM_EPOCHS = 2
+
+optimizer = optim.Adam(model.parameters(), lr=1e-3)
+
+
+def train(epoch):
+    batch_data_t = torch.FloatTensor(BATCH_SIZE, 784)
+    if cuda:
+        batch_data_t = batch_data_t.cuda()
+    batch_data = Variable(batch_data_t, requires_grad=False)
+    for i in range(0, training_data.size(0), BATCH_SIZE):
+        optimizer.zero_grad()
+        batch_data.data[:] = training_data[i:i + BATCH_SIZE]
+        recon_batch_data, mu, logvar = model(batch_data)
+        loss = loss_function(recon_batch_data, batch_data, mu, logvar)
+        loss.backward()
+        loss = loss.data[0]
+        optimizer.step()
+        if i % 10 == 0:
+            print('Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.4f}'.format(
+                epoch,
+                i + BATCH_SIZE, training_data.size(0),
+                float(i + BATCH_SIZE) / training_data.size(0) * 100,
+                loss / BATCH_SIZE))
+
+
+def test(epoch):
+    test_loss = 0
+    batch_data_t = torch.FloatTensor(TEST_BATCH_SIZE, 784)
+    if cuda:
+        batch_data_t = batch_data_t.cuda()
+    batch_data = Variable(batch_data_t, volatile=True)
+    for i in range(0, test_data.size(0), TEST_BATCH_SIZE):
+        print('Testing model: {}/{}'.format(i, test_data.size(0)), end='\r')
+        batch_data.data[:] = test_data[i:i + TEST_BATCH_SIZE]
+        recon_batch_data, mu, logvar = model(batch_data)
+        test_loss += loss_function(recon_batch_data, batch_data, mu, logvar)
+
+    test_loss = test_loss.data[0] / test_data.size(0)
+    print('TEST SET RESULTS:' + ' ' * 20)
+    print('Average loss: {:.4f}'.format(test_loss))
+
+
+for epoch in range(1, NUM_EPOCHS + 1):
+    train(epoch)
+    test(epoch)

VAE/requirements.txt

@@ -0,0 +1,3 @@
+torch
+tqdm
+six