X-Git-Url: https://www.fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=mine_mnist.py;h=5ab427fd7ec684f439cfa1c377e727968f2c3040;hb=1ef32d0b85a179a36a1cd7899b8301bcb8e563d2;hp=f573b894634e439134e64ac46f019472bb8aab41;hpb=5e18633709b82b0739e62d105b2685793be1d014;p=pytorch.git

diff --git a/mine_mnist.py b/mine_mnist.py
index f573b89..5ab427f 100755
--- a/mine_mnist.py
+++ b/mine_mnist.py
@@ -1,10 +1,5 @@
 #!/usr/bin/env python
 
-# @XREMOTE_HOST: elk.fleuret.org
-# @XREMOTE_EXEC: ~/conda/bin/python
-# @XREMOTE_PRE: ln -s ~/data/pytorch ./data
-# @XREMOTE_PRE: killall -q -9 python || echo "Nothing killed"
-
 import math, sys, torch, torchvision
 
 from torch import nn
@@ -12,26 +7,53 @@ from torch.nn import functional as F
 
 ######################################################################
 
-# Returns a pair of tensors (x, c), where x is a Nx2x28x28 containing
-# pairs of images of same classes (one per channel), and p is a 1d
-# long tensor with the count of pairs per class used
+train_set = torchvision.datasets.MNIST('./data/mnist/', train = True, download = True)
+train_input  = train_set.train_data.view(-1, 1, 28, 28).float()
+train_target = train_set.train_labels
+
+test_set = torchvision.datasets.MNIST('./data/mnist/', train = False, download = True)
+test_input = test_set.test_data.view(-1, 1, 28, 28).float()
+test_target = test_set.test_labels
+
+mu, std = train_input.mean(), train_input.std()
+train_input.sub_(mu).div_(std)
+test_input.sub_(mu).div_(std)
+
+used_MNIST_classes = torch.tensor([ 0, 1, 3, 5, 6, 7, 8, 9])
+# used_MNIST_classes = torch.tensor([ 0, 9, 7 ])
+# used_MNIST_classes = torch.tensor([ 3, 4, 7, 0 ])
+
+######################################################################
+
+# Returns a triplet of tensors (a, b, c), where a and b contain each
+# half of the samples, with a[i] and b[i] of same class for any i, and
+# c is a 1d long tensor with the count of pairs per class used.
+
+def create_MNIST_pair_set(train = False):
+    ua, ub = [], []
 
-def create_pair_set(used_classes, input, target):
-    u = []
+    if train:
+        input, target = train_input, train_target
+    else:
+        input, target = test_input, test_target
 
-    for i in used_classes:
+    for i in used_MNIST_classes:
         used_indices = torch.arange(input.size(0), device = target.device)\
                             .masked_select(target == i.item())
         x = input[used_indices]
         x = x[torch.randperm(x.size(0))]
-        # Careful with odd numbers of samples in a class
-        x = x[0:2 * (x.size(0) // 2)].reshape(-1, 2, 28, 28)
-        u.append(x)
+        hs = x.size(0)//2
+        ua.append(x.narrow(0, 0, hs))
+        ub.append(x.narrow(0, hs, hs))
 
-    x = torch.cat(u, 0).contiguous()
-    c = torch.tensor([x.size(0) for x in u])
+    a = torch.cat(ua, 0)
+    b = torch.cat(ub, 0)
+    perm = torch.randperm(a.size(0))
+    a = a[perm].contiguous()
+    b = b[perm].contiguous()
+    c = torch.tensor([x.size(0) for x in ua])
 
-    return x, c
+    return a, b, c
 
 ######################################################################
 
@@ -43,7 +65,9 @@ class Net(nn.Module):
         self.fc1 = nn.Linear(256, 200)
         self.fc2 = nn.Linear(200, 1)
 
-    def forward(self, x):
+    def forward(self, a, b):
+        # Make the two images a single two-channel image
+        x = torch.cat((a, b), 1)
         x = F.relu(F.max_pool2d(self.conv1(x), kernel_size = 3))
         x = F.relu(F.max_pool2d(self.conv2(x), kernel_size = 2))
         x = x.view(x.size(0), -1)
@@ -53,53 +77,69 @@ class Net(nn.Module):
 
 ######################################################################
 
-train_set = torchvision.datasets.MNIST('./data/mnist/', train = True, download = True)
-train_input  = train_set.train_data.view(-1, 1, 28, 28).float()
-train_target = train_set.train_labels
-
-mu, std = train_input.mean(), train_input.std()
-train_input.sub_(mu).div_(std)
-
-######################################################################
-
-# The information bound is the log of the number of classes in there
-
-# used_classes = torch.tensor([ 0, 1, 3, 5, 6, 7, 8, 9])
-used_classes = torch.tensor([ 3, 4, 7, 0 ])
-
 nb_epochs, batch_size = 50, 100
 
 model = Net()
+
+print('nb_parameters %d' % sum(x.numel() for x in model.parameters()))
+
 optimizer = torch.optim.Adam(model.parameters(), lr = 1e-3)
 
 if torch.cuda.is_available():
     model.cuda()
     train_input, train_target = train_input.cuda(), train_target.cuda()
+    test_input, test_target = test_input.cuda(), test_target.cuda()
 
 for e in range(nb_epochs):
-    input, count = create_pair_set(used_classes, train_input, train_target)
 
+    input_a, input_b, count = create_MNIST_pair_set(train = True)
+
+    # The information bound is the entropy of the class distribution
     class_proba = count.float()
     class_proba /= class_proba.sum()
     class_entropy = - (class_proba.log() * class_proba).sum().item()
 
-    input = input[torch.randperm(input.size(0))]
-    indep_input = input.clone()
-    indep_input[:, 1] = input[torch.randperm(input.size(0)), 1]
+    input_br = input_b[torch.randperm(input_b.size(0))]
 
-    mi = 0.0
+    acc_mi = 0.0
 
-    for batch, indep_batch in zip(input.split(batch_size), indep_input.split(batch_size)):
-        loss = - (model(batch).mean() - model(indep_batch).exp().mean().log())
-        mi -= loss.item()
+    for batch_a, batch_b, batch_br in zip(input_a.split(batch_size),
+                                          input_b.split(batch_size),
+                                          input_br.split(batch_size)):
+        mi = model(batch_a, batch_b).mean() - model(batch_a, batch_br).exp().mean().log()
+        loss = - mi
+        acc_mi += mi.item()
         optimizer.zero_grad()
         loss.backward()
         optimizer.step()
 
-    mi /= (input.size(0) // batch_size)
+    acc_mi /= (input_a.size(0) // batch_size)
 
-    print('%d %.04f %.04f'%(e, mi / math.log(2), class_entropy / math.log(2)))
+    print('%d %.04f %.04f' % (e, acc_mi / math.log(2), class_entropy / math.log(2)))
 
     sys.stdout.flush()
 
 ######################################################################
+
+input_a, input_b, count = create_MNIST_pair_set(train = False)
+
+for e in range(nb_epochs):
+    class_proba = count.float()
+    class_proba /= class_proba.sum()
+    class_entropy = - (class_proba.log() * class_proba).sum().item()
+
+    input_br = input_b[torch.randperm(input_b.size(0))]
+
+    acc_mi = 0.0
+
+    for batch_a, batch_b, batch_br in zip(input_a.split(batch_size),
+                                          input_b.split(batch_size),
+                                          input_br.split(batch_size)):
+        mi = model(batch_a, batch_b).mean() - model(batch_a, batch_br).exp().mean().log()
+        acc_mi += mi.item()
+
+    acc_mi /= (input_a.size(0) // batch_size)
+
+print('test %.04f %.04f'%(acc_mi / math.log(2), class_entropy / math.log(2)))
+
+######################################################################