Deep Learning with PyTorch

Name: Deep Learning with PyTorch
Rating: 4.1 (63 reviews)

Build useful and effective deep learning models with the PyTorch Deep Learning framework

Created byPackt Publishing

Last updated 5/2018

English

What you'll learn

Understand PyTorch and Deep Learning concepts
Build your neural network using Deep Learning techniques in PyTorch.
Perform basic operations on your dataset using tensors and variables
Build artificial neural networks in Python with GPU acceleration
See how CNN works in PyTorch with a simple computer vision example
Train your RNN model from scratch for text generation
Use Auto Encoders in PyTorch to remove noise from images
Perform reinforcement learning to solve OpenAI's Cartpole task
Extend your knowledge of Deep Learning by using PyTorch to solve your own machine learning problems

Course content

6 sections • 38 lectures • 4h 42m total length

The Course Overview6:29
This video provides an overview of the entire course.
Introduction to PyTorch6:18
Get introduced to PyTorch.

Develop a mental model of the PyTorch Deep Learning framework
Understand the salient features of PyTorch
Installing PyTorch on Linux and Windows10:41
Have your PyTorch based Linux and Windows environments ready.

Download the required software packages.
Setup the PyTorch based Linux environment.
Setup the PyTorch based Windows environment.
Installing CUDA4:41
Get your GPU based system CUDA enabled.

Know about CUDA
Download the required software
Install and test
Introduction to Tensors and Variables16:17
Familiarize yourself with Tensors and Variables.

Understand the role of a Tensor
Know what Variables are
Run PyTorch code to see them in action
Working with PyTorch and NumPy2:38
Familiarize yourself with the bridge between PyTorch and NumPy.

Understand the connection between PyTorch and NumPy
Run PyTorch code to see it in action
Working with PyTorch and GPU3:07
Execute computation on GPU.

Understand the PyTorch GPU model
See how data can be moved from CPU to GPU and back
Run PyTorch code to see it in action
Handling Datasets in PyTorch8:29
Access datasets in PyTorch.

Understand concepts like Dataset, Epoch, Batch, and Iteration
See how popular datasets can be accessed using TorchVision
Run PyTorch code to see it in action
Deep Learning Using PyTorch8:17
Get introduced to Deep Learning and know the PyTorch based projects we are going to execute in this course.

Understand the relation between Deep Learning and Machine Learning
Explore some use cases solved by Computer Vision and NLP
Get to know the topics and projects covered in rest of the course.

Building a Simple Neural Network13:26
Learn how to build a simple neural network in PyTorch.
Know about concepts like nodes, edges, weights and biases in a neuron
Understand the role of activation functions
Create a 3-layer neural network in PyTorch
Loss Functions in PyTorch2:07
Learn about the role of loss functions.

Understand the role of loss functions
Understand where loss functions fit in the training process
Know when to use Cross Entropy Loss
Optimizers in PyTorch3:50
Learn about optimizers.

Understand the role of optimizers
Understand where loss functions fit in the training process
Know about gradients and how the optimizer uses it
Training the Neural Network6:36
Learn how to train a model.

Know about the training loop
Understand the various operations within the training loop
Execute the training
Saving and Loading a Trained Neural Network1:27
Learn how to save models to disk and read them back.

Know what APIs to use in PyTorch to accomplish this
Training the Neural Network on a GPU3:46
Learn how to train your model on a GPU.

Know what changes to make in the model to accomplish this

Computer Vision Motivation4:57
Get motivated towards the subject.

Know about the practical applications of computer vision in the industry
Convolutional Neural Networks8:08
Understand Convolutional Neural Networks.

Get to know how images are represented and used
Understand the convolution operation
Understand the concept of pooling
The Convolution Operation9:28
Know how to load the MNIST dataset.

Learn about the MNIST dataset
Use torchvision to get the MNIST dataset
Create the DataLoader to iterate through the MNIST dataset
Concepts - Strides, Padding, and Pooling9:28
Learn some essential concepts around the convolution layers and the convolution operations.

Explain the concepts of strides, padding and pooling in the context of CNNs
Understand the effect of these on the activation maps
Loading and Using MNIST Dataset9:05
Learn how to load the MNIST

Learn about the MNIST dataset
Use torchvision to get the MNIST dataset
Create the DataLoader to iterate through the MNIST dataset
Building the Model8:57
Learn how to build CNN models

Use classes from PyTorch to construct the CNN model
Understand the dimensions of the activation map in each layer
Training and Testing11:53
See how to train the model and test it

Build the training loop to train the model
Visualize the loss and accuracy of the training and validation data
Learn how to do inferencing on a single item

Sequence Models Motivation4:55
Get motivated towards the subject.

Know about the practical applications of computer vision in the industry.
Word Embedding6:46
Understand the way text is represented in neural networks.

Learn about one hot vector and it’s short comings
Learn how words are represented as meaningful vectors of numbers
Learn about Embedding layer and GloVe and Word2Vec models
Recurrent Neural Networks10:44
Understand the neural network architecture to process sequential data.

Learn about the RNN architecture
Understand the various configurations in which RNNs can be used
Understand the problems fixed by LSTM and GRU cells
Building a Text Generation Model in PyTorch17:25
Create a model to generate Shakespeare like text.

Have intuitive understanding of language model
Training and Testing7:27
See how to train the model and test it.

Understand how to use the language model to generate text
Initialize and generate Shakespeare like text

Autoencoders Motivation4:32
Get motivated towards the subject.

Know about the practical applications of Autoencoders in the industry.
How Autoencoders Work3:21
Understand the technical details behind Autoencoders.

Learn about the typical architecture of Autoencoders
Understand what the bottleneck layer is
Understand how an Autoencoder is trained
Types of Autoencoders3:57
Learn about the most used Autoencoder variants.

Know the various types of Autoencoders
Learn about their unique characteristics
Building Denoising Autoencoder Using PyTorch11:22
Learn how to code an Autoencoder in PyTorch.

Learn the construction of an Autoencoder
Learn the way it needs to be trained
Training and Testing4:18
Learn how to train and use an Autoencoder.

Visualize the training and validation loss
Learn how to use a Denoising Autoencoder on a single image

Reinforcement Learning Motivation6:10
Get motivated towards the subject.

Know about the practical applications of reinforcement learning
Reinforcement Learning Concepts10:55
Understand the key concepts in reinforcement learning.

Learn the actors in reinforcement learning
Understand terms like state, action, reward, policy
Understand the Q-learning algorithm to train agents
DQN, Experience Replay6:07
Understand the Deep Q-Network algorithm and how to better train it.

Learn the construction of DQN
Understand how Experience replay helps train the network better
The OpenAI Gym Environment6:10
Get introduced to the OpenAI gym environment.

Know about the OpenAI project
See how to setup and initialize the environment
Building the Cartpole Agent Using DQN8:26
Build a reinforcement learning agent

Understand the various parts of the code to create the agent
Build everything using Python and PyTorch
Training and Testing9:51
Train and test run the agent.

Train the agent and monitor it’s progress
Run the agent standalone and see how it performs

Requirements

Python programming knowledge and minimal math skills (matrix/vector manipulation, simple probabilities) is assumed.

Description

This video course will get you up-and-running with one of the most cutting-edge deep learning libraries: PyTorch. Written in Python, PyTorch is grabbing the attention of all data science professionals due to its ease of use over other libraries and its use of dynamic computation graphs.

In this course, you will learn how to accomplish useful tasks using Convolutional Neural Networks to process spatial data such as images and using Recurrent Neural Networks to process sequential data such as texts. You will explore how you can make use of unlabeled data using Auto Encoders. You will also be training a neural network to learn how to balance a pole all by itself, using Reinforcement Learning. Throughout this journey, you will implement various mechanisms of the PyTorch framework to do these tasks.

By the end of the video course, you will have developed a good understanding of, and feeling for, the algorithms and techniques used. You'll have a good knowledge of how PyTorch works and how you can use it in to solve your daily machine learning problems.

This course uses Python 3.6, and PyTorch 0.3, while not the latest version available, it provides relevant and informative content for legacy users of Python, and PyTorch.

About the Author

Anand Saha is a software professional with 15 years' experience in developing enterprise products and services. Back in 2007, he worked with machine learning to predict call patterns at TATA Communications. At Symantec and Veritas, he worked on various features of an enterprise backup product used by Fortune 500 companies. Along the way he nurtured his interests in Deep Learning by attending Coursera and Udacity MOOCs.

He is passionate about Deep Learning and its applications; so much so that he quit Veritas at the beginning of 2017 to focus full time on Deep Learning practices. Anand built pipelines to detect and count endangered species from aerial images, trained a robotic arm to pick and place objects, and implemented NIPS papers. His interests lie in computer vision and model optimization.

Who this course is for:

This course is for Python programmers who have some knowledge of machine learning and want to build Deep Learning systems with PyTorch.

Deep Learning with PyTorch

What you'll learn

Explore related topics

Course content

Getting Started With PyTorch9 lectures • 1hr 7min

Training Your First Neural Network6 lectures • 31min

Computer Vision – CNN for Digits Recognition7 lectures • 1hr 2min

Sequence Models – RNN for Text Generation5 lectures • 47min

Autoencoder - Denoising Images5 lectures • 28min

Reinforcement Learning – Balance Cartpole Using DQN6 lectures • 48min

Requirements

Description

Who this course is for: