## AI & Deep Learning with TensorFlow

- 16k Enrolled Learners
- Weekend
- Live Class

Let’s start this **PyTorch Tutorial** blog by establishing a fact that **Deep Learning** is something that is being used by **everyone** today, ranging from **Virtual Assistance** to getting **recommendations** while **shopping**! With newer tools emerging to make better use of Deep Learning, programming and implementation have become **easier**.

This **PyTorch Tutorial** will give you a **complete insight** into PyTorch in the following sequence:

- What is PyTorch
- Features of PyTorch
- Installing PyTorch
- The NumPy Bridge
- PyTorch: AutoGrad Module
- Use Case: Image Classifier

Python is **preferred** for coding and working with Deep Learning and hence has a **wide spectrum** of frameworks to choose from. Such as:

- TensorFlow
- PyTorch
- Keras
- Theano
- Lasagne

It’s a **Python** based scientific computing **package** targeted at two sets of audiences:

- A
**replacement**for NumPy to make use of the power of**GPUs**. - Deep Learning
**research platform**that provides maximum**flexibility**and**speed**.

**Native support**for Python and use of its libraries

- Actively used in the
**development of Facebook**for all of it’s Deep Learning requirements in the platform. - PyTorch ensures an
**easy to use API**which helps with easier usability and better understanding when making use of the API. **Dynamic Computation Graphs**are a major highlight here as they ensure the graph build-up dynamically – at every point of code execution, the graph is built along and can be manipulated at run-time.**PyTorch is fast**and**feels native**, hence ensuring easy coding and fast processing.**The support for CUDA**ensures that the code can run on the GPU, thereby decreasing the time needed to run the code and increasing the overall performance of the system.

Moving ahead in this PyTorch Tutorial, let’s see how simple it is to actually install PyTorch on your machine.

It’s pretty **straight-forward** based on the system properties such as the **Operating System** or the package managers. It can be installed from the **Command Prompt** or within an **IDE** such as **PyCharm** etc.

Next up on this PyTorch Tutorial blog, let us check out how **NumPy** is integrated into PyTorch.

Tensors are similar to NumPy’s n dimensional arrays, with the addition being that Tensors can also be used on a GPU to accelerate computing.

Let’s construct a simple tensor and check the output. First let’s check out on how we can construct a 5×3 matrix which is uninitiated:

x = torch.empty(5, 3) print(x)

Output:

tensor([[8.3665e+22, 4.5580e-41, 1.6025e-03], [3.0763e-41, 0.0000e+00, 0.0000e+00], [0.0000e+00, 0.0000e+00, 3.4438e-41], [0.0000e+00, 4.8901e-36, 2.8026e-45], [6.6121e+31, 0.0000e+00, 9.1084e-44]])

Now let’s construct a randomly initialized matrix:

x = torch.rand(5, 3) print(x)

Output:

tensor([[0.1607, 0.0298, 0.7555], [0.8887, 0.1625, 0.6643], [0.7328, 0.5419, 0.6686], [0.0793, 0.1133, 0.5956], [0.3149, 0.9995, 0.6372]])

Construct a tensor directly from data:

```
x = torch.tensor([5.5, 3])
print(x)
```

Output:

`tensor([5.5000, 3.0000])`

**Tensor Operations **

There are multiple syntaxes for operations. In the following example, we will take a look at the addition operation:

```
y = torch.rand(5, 3)
print(x + y)
```

Output:

```
tensor([[ 0.2349, -0.0427, -0.5053],
[ 0.6455, 0.1199, 0.4239],
[ 0.1279, 0.1105, 1.4637],
[ 0.4259, -0.0763, -0.9671],
[ 0.6856, 0.5047, 0.4250]])
```

Resizing: If you want to reshape/resize a tensor, you can use “torch.view”:

```
x = torch.randn(4, 4)
y = x.view(16)
z = x.view(-1, 8) # the size -1 is inferred from other dimensions
print(x.size(), y.size(), z.size())
```

Output:

`torch.Size([4, 4]) torch.Size([16]) torch.Size([2, 8])`

**NumPy is a library** for the Python programming language, adding support for large, multi-dimensional** arrays** and **matrices**, along with a large collection of** high-level mathematical functions** to operate on these arrays.

It is also used as:

**Library**providing**tools**for integrating C/C++ and FORTRAN code.- Operations with
**linear algebra**,**Fourier transforms**and**random number**capabilities.

Besides its obvious scientific uses, NumPy can also be used as an efficient **multi-dimensional container** of generic data and arbitrary data-types can be defined as well.

This allows **NumPy** to **seamlessly** and speedily **integrate** with a wide variety of **databases!**

**Converting** a Torch Tensor to a NumPy array and vice versa** is a breeze!**

The Torch Tensor and NumPy array will **share their underlying memory locations** and changing one will change the other.

a = torch.ones(5) print(a)

```
Output: tensor([1., 1., 1., 1., 1.])
```

b = a.numpy() print(b)

`Output: [1. 1. 1. 1. 1.]`

Let’s perform a **sum operation** and check the **changes** in the values:

a.add_(1) print(a) print(b)

`Output: tensor([2., 2., 2., 2., 2.]) [2. 2. 2. 2. 2.]`

import numpy as no a = np.ones(5) b = torch.from_numpy(a) np.add(a, 1, out=a) print(a) print(b)

Output:

```
[2. 2. 2. 2. 2.]
tensor([2., 2., 2., 2., 2.], dtype=torch.float64)
```

So, as you can see, it is as **simple** as that!

Next up on this PyTorch Tutorial blog, let’s check out the **AutoGrad module** of PyTorch.

The **autograd** package provides **automatic differentiation** for all operations on Tensors.

It is a **define-by-run framework**, which means that your backprop is defined by how your code is run, and that every single **iteration** can be **different**.

Next up on this PyTorch Tutorial Blog, let’s look an interesting and a simple use case.

Generally, when you have to deal with image, text, audio or video data, you can use **standard python packages** that load data into a **Numpy** array. Then you can convert this array into a **torch.*Tensor**.

- For
**images**, packages such as**Pillow**and**OpenCV**are useful. - For
**audio**, packages such as**Scipy**and**Librosa**. - For
**text**, either raw Python,**Cython**based loading or**NLTK**and**SpaCy**are useful.

Specifically for vision, there is a package called **torchvision**, that has **data loaders** for common datasets such as **Imagenet, CIFAR10, MNIST, etc**. and data transformers for images.

This provides a **huge convenience** and **avoids writing boilerplate code.**

For this tutorial, we will use the **CIFAR10** dataset.

It has the **classes**: ‘airplane’, ‘automobile’, ‘bird’, ‘cat’, ‘deer’, ‘dog’, ‘frog’, ‘horse’, ‘ship’, ‘truck’. The images in CIFAR-10 are of size 3x32x32, i.e. 3-channel color images of 32×32 pixels in size as shown below:

We will do the following steps in order:

Using **torchvision**, it is **very easy** to load CIFAR10!

It is as simple as follows:

import torch import torchvision import torchvision.transforms as transforms

The output of torchvision datasets are **PILImage** images of range [0, 1]. We transform them to Tensors of normalized range [-1, 1].

transform = transforms.Compose( [transforms.ToTensor(), transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))]) trainset = torchvision.datasets.CIFAR10(root='./data', train=True, download=True, transform=transform) trainloader = torch.utils.data.DataLoader(trainset, batch_size=4, shuffle=True, num_workers=2) testset = torchvision.datasets.CIFAR10(root='./data', train=False, download=True, transform=transform) testloader = torch.utils.data.DataLoader(testset, batch_size=4, shuffle=False, num_workers=2) classes = ('plane', 'car', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck')

Output:

`Downloading https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz to ./data/cifar-10-python.tar.gz Files already downloaded and verified `

Next, let us print some **training images** from the** dataset!**

import matplotlib.pyplot as plt import numpy as np # functions to show an image def imshow(img): img = img / 2 + 0.5 # unnormalize npimg = img.numpy() plt.imshow(np.transpose(npimg, (1, 2, 0))) # get some random training images dataiter = iter(trainloader) images, labels = dataiter.next() # show images imshow(torchvision.utils.make_grid(images)) # print labels print(' '.join('%5s' % classes[labels[j]] for j in range(4)))

Output:

`dog bird horse horse`

Consider the case to use** 3-channel images** (Red, Green and Blue). Here’s the** code** to define the architecture of the CNN:

```
import torch.nn as nn
import torch.nn.functional as F
class Net(nn.Module):
def __init__(self):
super(Net, self).__init__()
self.conv1 = nn.Conv2d(3, 6, 5)
self.pool = nn.MaxPool2d(2, 2)
self.conv2 = nn.Conv2d(6, 16, 5)
self.fc1 = nn.Linear(16 * 5 * 5, 120)
self.fc2 = nn.Linear(120, 84)
self.fc3 = nn.Linear(84, 10)
def forward(self, x):
x = self.pool(F.relu(self.conv1(x)))
x = self.pool(F.relu(self.conv2(x)))
x = x.view(-1, 16 * 5 * 5)
x = F.relu(self.fc1(x))
x = F.relu(self.fc2(x))
x = self.fc3(x)
return x
net = Net()
```

We will need to **define** the loss function. In this case we can make use of a **Classification Cross-Entropy** loss. We’ll also be using **SGD** with **momentum** as well.

Basically, the Cross-Entropy Loss is a probability value ranging from 0-1. The **perfect model** will a Cross Entropy Loss of **0** but it might so happen that the **expected value** may be 0.2 but you are getting 2. This will lead to a **very high loss** and not be efficient at all!

import torch.optim as optim criterion = nn.CrossEntropyLoss() optimizer = optim.SGD(net.parameters(), lr=0.001, momentum=0.9)

This is when things start to get interesting! We simply have to **loop** over our **data iterator**, and **feed** the inputs to the **network** and **optimize**.

for epoch in range(2): # loop over the dataset multiple times running_loss = 0.0 for i, data in enumerate(trainloader, 0): # get the inputs inputs, labels = data # zero the parameter gradients optimizer.zero_grad() # forward + backward + optimize outputs = net(inputs) loss = criterion(outputs, labels) loss.backward() optimizer.step() # print statistics running_loss += loss.item() if i % 2000 == 1999: # print every 2000 mini-batches print('[%d, %5d] loss: %.3f' % (epoch + 1, i + 1, running_loss / 2000)) running_loss = 0.0 print('Finished Training')

Output:

```
[1, 2000] loss: 2.236
[1, 4000] loss: 1.880
[1, 6000] loss: 1.676
[1, 8000] loss: 1.586
[1, 10000] loss: 1.515
[1, 12000] loss: 1.464
[2, 2000] loss: 1.410
[2, 4000] loss: 1.360
[2, 6000] loss: 1.360
[2, 8000] loss: 1.325
[2, 10000] loss: 1.312
[2, 12000] loss: 1.302
Finished Training
```

We have trained the network for **2 passes** over the **training dataset**. But we need to **check** if the network has learnt anything at all.

We will check this by **predicting the class labe**l that the neural network outputs, and **checking it against the ground-truth.** If the prediction is **correct**, we **add** the sample to the list of correct predictions.

Okay, first step! Let us display an image from the test set to get familiar.

dataiter = iter(testloader) images, labels = dataiter.next() # print images imshow(torchvision.utils.make_grid(images)) print('GroundTruth: ', ' '.join('%5s' % classes[labels[j]] for j in range(4)))

Output:

```
GroundTruth: cat ship ship plane
```

Okay, now let us see what the Neural Network thinks these examples above are:

```
outputs = net(images)
```

The outputs are **energies** for the 10 classes. Higher the energy for a class, the more the network thinks that the image is of the **particular class**. So, let’s get the index of the **highest energy**:

```
predicted = torch.max(outputs, 1)
print('Predicted: ', ' '.join('%5s' % classes[predicted[j]]
for j in range(4)))
```

Output:

`Predicted: cat car car plane`

**The results seem pretty good.**

Next up on this PyTorch Tutorial blog, let us look at how the network performs on the whole dataset!

```
correct = 0
total = 0
with torch.no_grad():
for data in testloader:
images, labels = data
outputs = net(images)
_, predicted = torch.max(outputs.data, 1)
total += labels.size(0)
correct += (predicted == labels).sum().item()
print('Accuracy of the network on the 10000 test images: %d %%' % (
100 * correct / total))
```

Output:

```
Accuracy of the network on the 10000 test images: 54 %
```

That looks** better than chance**, which is **10%** accuracy (randomly picking a class out of 10 classes).

**Seems like the network learned something!**

What are the classes that performed **well**, and the classes that **did not** perform well?

```
class_correct = list(0. for i in range(10))
class_total = list(0. for i in range(10))
with torch.no_grad():
for data in testloader:
images, labels = data
outputs = net(images)
_, predicted = torch.max(outputs, 1)
c = (predicted == labels).squeeze()
for i in range(4):
label = labels[i]
class_correct[label] += c[i].item()
class_total[label] += 1
for i in range(10):
print('Accuracy of %5s : %2d %%' % (
classes[i], 100 * class_correct[i] / class_total[i]))
```

Output:

```
Accuracy of plane : 61 %
Accuracy of car : 85 %
Accuracy of bird : 46 %
Accuracy of cat : 23 %
Accuracy of deer : 40 %
Accuracy of dog : 36 %
Accuracy of frog : 80 %
Accuracy of horse : 59 %
Accuracy of ship : 65 %
Accuracy of truck : 46 %
```

In this **PyTorch Tutorial** blog, we made sure to **train** a small **Neural Network** which **classifies** images and it turned out perfectly as expected!

**Check out these interesting blogs on the following topics:**

**Artificial Intelligence with Deep Learning!****TensorFlow Tutorial****Neural Network Tutorial****Backpropagation Tutorial**