UNIT IV
Convolutional Neural Networks: Nerual Network and Representation Learing, Convolutional Layers,
Multichannel Convolution Operation, Recurrent Neural Networks: Introduction to RNN, RNN Code,
PyTorch Tensors: Deep Learning with PyTorch, CNN in PyTorch
……………………………………………………………………………………………………………………………
Convolutional neural networks (CNNs) are a type of deep learning neural network that is specifically designed
for image processing and computer vision tasks. CNNs are inspired by the structure and function of the human
visual cortex, which is the part of the brain that is responsible for processing visual information.
CNNs have a number of advantages over other types of neural networks for image processing and computer
vision tasks:
• They are able to extract features from images that are invariant to translation, rotation, and scaling.
This means that the same features can be detected even if the object in the image is in a different
position or orientation.
• They are able to extract features from images that are hierarchically organized. This means that they
can start by extracting simple features, such as edges and corners, and then use those features to
extract more complex features, such as faces and objects.
• They are computationally efficient. This is because the convolution operation at the heart of CNNs
can be implemented using the fast Fourier transform (FFT).
CNNs have revolutionized the field of computer vision. They are now used in a wide range of applications,
including:
• Image classification: CNNs can be used to classify images into different categories, such as cats, dogs,
and cars.
• Object detection: CNNs can be used to detect objects in images, such as pedestrians, cars, and traffic
signs.
• Facial recognition: CNNs can be used to recognize faces in images.
• Medical imaging: CNNs can be used to analyze medical images, such as X-rays and MRI scans, to
diagnose diseases and identify abnormalities.
• Natural language processing: CNNs can be used to extract features from text, which can then be used
for tasks such as sentiment analysis and machine translation.
CNNs are a powerful tool for a wide range of applications. They are still under active development, and
new applications for CNNs are being discovered all the time.
Here are some specific examples of how CNNs are being used today:
• Facebook uses CNNs to recognize faces in photos.
• Google uses CNNs to power its image search engine.
• Tesla uses CNNs to power its self-driving cars.
• Doctors use CNNs to analyze medical images and diagnose diseases.
• Researchers are using CNNs to develop new methods for machine translation and text analysis.
CNNs are a powerful and versatile tool that is transforming the way we interact with the world around us.
1 | ©www.tutorialtpint.net – Prepared By D.Venkata Reddy M.Tech(Ph.D), UGC NET, AP SET Qualified
,Artificial neural networks (ANNs) Vs convolutional neural networks (CNNs)
Artificial neural networks (ANNs) and convolutional neural networks (CNNs) are both types of deep
learning neural networks. However, they have different architectures and are used for different types of
tasks.
ANNs are general-purpose neural networks that can be used for a variety of tasks, including classification,
regression, and clustering. They are typically made up of a series of fully connected layers, meaning that each
neuron in one layer is connected to every neuron in the next layer.
CNNs are a type of ANN that is specifically designed for image processing and computer vision tasks. They
are made up of a series of convolutional layers, which are able to extract features from images that are
invariant to translation, rotation, and scaling.
Here is a table that summarizes the key differences between ANNs and CNNs:
Characteristic ANN CNN
Architecture Fully connected layers Convolutional layers
Image processing, computer
Applications General-purpose
vision
Able to extract invariant features
Advantages Flexible and versatile
from images
Can be computationally Requires a large amount of
Disadvantages
expensive labeled training data
Which type of neural network to use depends on the specific task at hand. If you are working on a general-
purpose task, such as classification or regression, then an ANN may be a good choice. If you are working on
an image processing or computer vision task, then a CNN is likely to be a better choice.
Here are some examples of when to use ANNs and CNNs:
• ANNs:
o Classifying text documents into different categories
o Predicting customer churn
o Recommending products to customers
• CNNs:
o Classifying images of objects
o Detecting objects in images
o Segmenting images
2 | ©www.tutorialtpint.net – Prepared By D.Venkata Reddy M.Tech(Ph.D), UGC NET, AP SET Qualified
, 1. Nerual Network and Representation Learing
Neural networks initially receive data on observations, with each observation represented by some
number n features.
A simple neural network model with one hidden layer performed better than a model without that
hidden layer.
One reason is that the neural network could learn nonlinear relationships between input and output.
However, a more general reason is that in machine learning, we often need linear combinations of our
original features in order to effectively predict our target.
Let's say that the pixel values for an MNIST digit are x1 through x784.
There may be many other such combinations, all of which contribute positively or negatively to the
probability that an image is of a particular digit.
Neural networks can automatically discover combinations of the original features that are important
through their training process.
This process of learning which combinations of features are important is known as representation
learning, and it's the main reason why neural networks are successful across different domains.
Is there any reason to modify this process for image data?
The fundamental insight that suggests the answer is "yes" is that in images, the interesting "combinations of
features" (pixels) tend to come from pixels that are close together in the image.
We want to exploit this fundamental fact about image data: that the order of the features matters since it tells
us which pixels are near each other spatially.
But how do we do it?
1.1. A Different Architecture for Image Data
The solution, at a high level, will be to create combinations of features, as before, but an order of magnitude
more of them, and have each one be only a combination of the pixels from a small rectangular patch in the
input image. Figure 5-2 describes this.
3 | ©www.tutorialtpint.net – Prepared By D.Venkata Reddy M.Tech(Ph.D), UGC NET, AP SET Qualified
Convolutional Neural Networks: Nerual Network and Representation Learing, Convolutional Layers,
Multichannel Convolution Operation, Recurrent Neural Networks: Introduction to RNN, RNN Code,
PyTorch Tensors: Deep Learning with PyTorch, CNN in PyTorch
……………………………………………………………………………………………………………………………
Convolutional neural networks (CNNs) are a type of deep learning neural network that is specifically designed
for image processing and computer vision tasks. CNNs are inspired by the structure and function of the human
visual cortex, which is the part of the brain that is responsible for processing visual information.
CNNs have a number of advantages over other types of neural networks for image processing and computer
vision tasks:
• They are able to extract features from images that are invariant to translation, rotation, and scaling.
This means that the same features can be detected even if the object in the image is in a different
position or orientation.
• They are able to extract features from images that are hierarchically organized. This means that they
can start by extracting simple features, such as edges and corners, and then use those features to
extract more complex features, such as faces and objects.
• They are computationally efficient. This is because the convolution operation at the heart of CNNs
can be implemented using the fast Fourier transform (FFT).
CNNs have revolutionized the field of computer vision. They are now used in a wide range of applications,
including:
• Image classification: CNNs can be used to classify images into different categories, such as cats, dogs,
and cars.
• Object detection: CNNs can be used to detect objects in images, such as pedestrians, cars, and traffic
signs.
• Facial recognition: CNNs can be used to recognize faces in images.
• Medical imaging: CNNs can be used to analyze medical images, such as X-rays and MRI scans, to
diagnose diseases and identify abnormalities.
• Natural language processing: CNNs can be used to extract features from text, which can then be used
for tasks such as sentiment analysis and machine translation.
CNNs are a powerful tool for a wide range of applications. They are still under active development, and
new applications for CNNs are being discovered all the time.
Here are some specific examples of how CNNs are being used today:
• Facebook uses CNNs to recognize faces in photos.
• Google uses CNNs to power its image search engine.
• Tesla uses CNNs to power its self-driving cars.
• Doctors use CNNs to analyze medical images and diagnose diseases.
• Researchers are using CNNs to develop new methods for machine translation and text analysis.
CNNs are a powerful and versatile tool that is transforming the way we interact with the world around us.
1 | ©www.tutorialtpint.net – Prepared By D.Venkata Reddy M.Tech(Ph.D), UGC NET, AP SET Qualified
,Artificial neural networks (ANNs) Vs convolutional neural networks (CNNs)
Artificial neural networks (ANNs) and convolutional neural networks (CNNs) are both types of deep
learning neural networks. However, they have different architectures and are used for different types of
tasks.
ANNs are general-purpose neural networks that can be used for a variety of tasks, including classification,
regression, and clustering. They are typically made up of a series of fully connected layers, meaning that each
neuron in one layer is connected to every neuron in the next layer.
CNNs are a type of ANN that is specifically designed for image processing and computer vision tasks. They
are made up of a series of convolutional layers, which are able to extract features from images that are
invariant to translation, rotation, and scaling.
Here is a table that summarizes the key differences between ANNs and CNNs:
Characteristic ANN CNN
Architecture Fully connected layers Convolutional layers
Image processing, computer
Applications General-purpose
vision
Able to extract invariant features
Advantages Flexible and versatile
from images
Can be computationally Requires a large amount of
Disadvantages
expensive labeled training data
Which type of neural network to use depends on the specific task at hand. If you are working on a general-
purpose task, such as classification or regression, then an ANN may be a good choice. If you are working on
an image processing or computer vision task, then a CNN is likely to be a better choice.
Here are some examples of when to use ANNs and CNNs:
• ANNs:
o Classifying text documents into different categories
o Predicting customer churn
o Recommending products to customers
• CNNs:
o Classifying images of objects
o Detecting objects in images
o Segmenting images
2 | ©www.tutorialtpint.net – Prepared By D.Venkata Reddy M.Tech(Ph.D), UGC NET, AP SET Qualified
, 1. Nerual Network and Representation Learing
Neural networks initially receive data on observations, with each observation represented by some
number n features.
A simple neural network model with one hidden layer performed better than a model without that
hidden layer.
One reason is that the neural network could learn nonlinear relationships between input and output.
However, a more general reason is that in machine learning, we often need linear combinations of our
original features in order to effectively predict our target.
Let's say that the pixel values for an MNIST digit are x1 through x784.
There may be many other such combinations, all of which contribute positively or negatively to the
probability that an image is of a particular digit.
Neural networks can automatically discover combinations of the original features that are important
through their training process.
This process of learning which combinations of features are important is known as representation
learning, and it's the main reason why neural networks are successful across different domains.
Is there any reason to modify this process for image data?
The fundamental insight that suggests the answer is "yes" is that in images, the interesting "combinations of
features" (pixels) tend to come from pixels that are close together in the image.
We want to exploit this fundamental fact about image data: that the order of the features matters since it tells
us which pixels are near each other spatially.
But how do we do it?
1.1. A Different Architecture for Image Data
The solution, at a high level, will be to create combinations of features, as before, but an order of magnitude
more of them, and have each one be only a combination of the pixels from a small rectangular patch in the
input image. Figure 5-2 describes this.
3 | ©www.tutorialtpint.net – Prepared By D.Venkata Reddy M.Tech(Ph.D), UGC NET, AP SET Qualified
