Convolutional Neural Network using Pytorch(Fashion-MNIST)

vanilla_cnn_pytorch.py

from __future__ import print_function
import torch
import torchvision
import torchvision.transforms as transforms
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim

import numpy as np
import matplotlib.pyplot as plt
torch.set_printoptions(linewidth = 120)

from sklearn.metrics import confusion_matrix
from sklearn.utils.multiclass import unique_labels
from sklearn.metrics import accuracy_score
from sklearn.metrics import f1_score

train_set = torchvision.datasets.FashionMNIST(
        root = './data/FashionMNIST',
        train = True,
        download = True,
        transform = transforms.Compose([
                transforms.ToTensor()
                ])
        )

train_loader = torch.utils.data.DataLoader(train_set, batch_size= 10)
# no of training sample
print(len(train_set))
# training labels
print(train_set.train_labels)
#frequency of label in training sample
print(train_set.train_labels.bincount())

sample = next(iter(train_set))
#length of sample
print(len(sample))
sample_image, sample_label = sample

# image size
print(sample_image.shape)

plt.imshow(sample_image.squeeze(), cmap='gray')
print('label:', sample_label)

batch = next(iter(train_loader))
images, labels = batch

print(images.shape)

#display 10 images in batch

grid = torchvision.utils.make_grid(images, nrow = 10)
plt.figure(figsize= (15,15))
plt.imshow(np.transpose(grid, (1,2,0)))

#build neural network
class FMnistNetwork(nn.Module):
    def __init__(self):
        super(FMnistNetwork, self).__init__()
        self.conv1 = nn.Conv2d(in_channels = 1, out_channels=6, kernel_size = 5)
        self.conv2 = nn.Conv2d(in_channels = 6, out_channels=12, kernel_size=5)
        
        self.fc1 = nn.Linear(in_features=12*4*4, out_features= 120)
        self.fc2 = nn.Linear(in_features = 120, out_features = 60)
        self.out = nn.Linear(in_features= 60, out_features = 10)
        
        
    def forward(self, tensor):
        
        # hidden layer 1
        tensor = self.conv1(tensor)
        tensor = F.relu(tensor)
        tensor = F.max_pool2d(tensor, kernel_size = 2, stride= 2)
        
        # hidden layer 2
        
        tensor = self.conv2(tensor)
        tensor = F.relu(tensor)
        tensor = F.max_pool2d(tensor, kernel_size = 2, stride = 2)
        
        #hidden layer 3
        
        tensor = tensor.reshape(-1, 12 * 4* 4)
        tensor = self.fc1(tensor)
        tensor = F.relu(tensor)
        
        #hidden layer 4
        
        tensor = self.fc2(tensor)
        tensor = F.relu(tensor)
        
        #output layer
        
        tensor = self.out(tensor)
        
        return tensor
    

# test network with one image

torch.set_grad_enabled(False)
fnn_dry_test_nn = FMnistNetwork()
print(fnn_dry_test_nn)

dry_test_sample = next(iter(train_set))
dry_test_image, dry_test_label = dry_test_sample

print(dry_test_image.shape)
#change dry_test_image rank as Batch, Channel, H, W

dry_test_image = dry_test_image.unsqueeze(0)
print(dry_test_image.shape)

dry_test_pred = fnn_dry_test_nn(dry_test_image)
print(dry_test_pred)
print(dry_test_label)
print(dry_test_pred.argmax(dim = 1)) 


# Output size formula
# n x n input
# f x f filter/kernel
# p is padding
# s is stride

# then output size will be (n - f + 2p)/s + 1

torch.set_grad_enabled(True)

fnn = FMnistNetwork()
optimizer = optim.Adam(fnn.parameters(), lr=0.001)
train_loader = torch.utils.data.DataLoader(train_set, batch_size= 100)
total_loss = 0
for epoch in range(5):
    
    for batch in train_loader:
        images, labels = batch
        
        preds = fnn(images)
        # calculate loss
        loss = F.cross_entropy(preds, labels)
        optimizer.zero_grad()
        # calculate gradients
        loss.backward()
        # update weights
        optimizer.step()
        total_loss += loss.item()
        
    print("epoch:", epoch, "loss:", total_loss)


@torch.no_grad()
def get_all_prediction(model, loader):
    preds = torch.tensor([])
    for batch in loader:
        images, labels = batch
        batch_predictions = model(images)
        preds = torch.cat((preds, batch_predictions), dim = 0)
    return preds


def plot_confusion_matrix(y_true, y_pred, classes,
                          normalize=False,
                          title=None,
                          cmap=plt.cm.Blues):
    """
    This function prints and plots the confusion matrix.
    Normalization can be applied by setting `normalize=True`.
    """
    if not title:
        if normalize:
            title = 'Normalized confusion matrix'
        else:
            title = 'Confusion matrix, without normalization'

    # Compute confusion matrix
    cm = confusion_matrix(y_true, y_pred)
    # Only use the labels that appear in the data
    #classes = classes[unique_labels(y_true, y_pred)]
    if normalize:
        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]
        print("Normalized confusion matrix")
    else:
        print('Confusion matrix, without normalization')

    print(cm)

    fig, ax = plt.subplots()
    im = ax.imshow(cm, interpolation='nearest', cmap=cmap)
    ax.figure.colorbar(im, ax=ax)
    # We want to show all ticks...
    ax.set(xticks=np.arange(cm.shape[1]),
           yticks=np.arange(cm.shape[0]),
           # ... and label them with the respective list entries
           xticklabels=classes, yticklabels=classes,
           title=title,
           ylabel='True label',
           xlabel='Predicted label')

    # Rotate the tick labels and set their alignment.
    plt.setp(ax.get_xticklabels(), rotation=45, ha="right",
             rotation_mode="anchor")

    # Loop over data dimensions and create text annotations.
    fmt = '.2f' if normalize else 'd'
    thresh = cm.max() / 2.
    for i in range(cm.shape[0]):
        for j in range(cm.shape[1]):
            ax.text(j, i, format(cm[i, j], fmt),
                    ha="center", va="center",
                    color="white" if cm[i, j] > thresh else "black")
    fig.tight_layout()
    return ax


np.set_printoptions(precision=2)

train_preds = get_all_prediction(fnn, train_loader)
label_dict = {
 'T-shirt/top',
 'Trouser',
 'Pullover',
 'Dress',
 'Coat',
 'Sandal',
 'Shirt',
 'Sneaker',
 'Bag',
 'Ankle boot',
}


# Plot non-normalized confusion matrix


plt.figure(figsize=(50,50))
plot_confusion_matrix(train_set.targets, train_preds.argmax(dim=1), classes=label_dict,
                      title='Confusion matrix')

plt.show()

print('Accuracy:', accuracy_score(train_set.targets, train_preds.argmax(dim=1)))
print('F1:', f1_score(train_set.targets, train_preds.argmax(dim=1), average='weighted'))