Description
Chapter 1 - Introduction Deep Learning
A brief introduction to Machine Learning and Deep Learning. We explore foundational topics within the subject that provide us the building blocks for several topics within the subject.
Chapter 2 - Introduction to PyTorch
A quick-start guide to PyTorch and a comprehensive introduction to tensors, linear algebra and mathematical operations for Tensors. The chapter provides the required PyTorch foundations for readers to meaningfully implement practical Deep Learning solutions for various topics within the book. Advanced PyTorch topics are explored as and when touch-based during the course of exercises in later chapter.
Chapter 3- Feed Forward Networks (30 Pages)
In this chapter, we explore the building blocks of a neural network and build an intuition on training and evaluating networks. We briefly explore loss functions, activation functions, optimizers, backpropagation, that could be used for training. Finally, we would stitch together each of these smaller components into a full-fledged feed-forward neural network with PyTorch.
Chapter 4-Automatic Differentiation in Deep Learning
In this chapter we open this black box topic within backpropagation that enables training of neural networks i.e. automatic differentiation. We cover a brief history of other techniques that were ruled out in favor of automatic differentiation and study the topic with a practical example and implement the same using PyTorchs Autograd module.
Chapter 5 - Training Deep Neural Networks
In this chapter we explore few additional important topics around deep learning and implement them into a practical example. We will delve into specifics of model performance and study in detail about overfitting and underfitting, hyperparameter tuning and regularization. Finally, we will leverage a real dataset and combined our learnings from the beginning of this book into a practical example using PyTorch.
Chapter 6 - Convolutional Neural Networks (35 Pages)
Introduction to Convolutional Neural Networks for Computer Vision. We explore the core components with CNNs with examples to understand the internals of the network, build an intuition around the automated feature extraction, parameter sharing and thus understand the holistic process of training CNNs with incremental building blocks. We also leverage hands-on exercises to study the practical implementation of CNNs for a simple dataset i.e. MNIST (classification of handwritten digits), and later extend the exercise for a binary classification use-case with the popular cats and dogs' dataset.Chapter 7 - Recurrent Neural Networks
Introduction to Recurrent Neural Networks and its variants (viz. Bidirectional RNNs and LSTMs). We explore the construction of a recurrent unit, study the mathematical background and build intuition around how RNNs are trained by exploring a simple four step unrolled network. We then explore hands-on exercises in natural language processing that leverages vanilla RNNs and later improve their performance by using Bidirectional RNNS combined with LSTM layers.
Chapter 8 - Recent advances in Deep Learning
A brief note of the cutting-edge advancements in the field will be added. We explore important inventions within the field with no implementation details, however focus on the applications and the path forward.Author: Nikhil Ketkar, Jojo Moolayil
Publisher: Apress
Published: 04/10/2021
Pages: 306
Binding Type: Paperback
Weight: 1.01lbs
Size: 9.21h x 6.14w x 0.68d
ISBN13: 9781484253632
ISBN10: 1484253639
BISAC Categories:
- Computers | Computer Science
- Computers | Programming | Object Oriented
- Mathematics | Logic
About the Author
Nikhil S. Ketkar currently leads the Machine Learning Platform team at Flipkart, India's largest e-commerce company. He received his Ph.D. from Washington State University. Following that he conducted postdoctoral research at University of North Carolina at Charlotte, which was followed by a brief stint in high frequency trading at Transmaket in Chicago. More recently he led the data mining team in Guavus, a startup doing big data analytics in the telecom domain and Indix, a startup doing data science in the e-commerce domain. His research interests include machine learning and graph theory.
Jojo Moolayil is an artificial intelligence, deep learning, machine learning, and decision science professional with over five years of industrial experience and is a published author of the book Smarter Decisions - The Intersection of IoT and Decision Science. He has worked with several industry leaders on high-impact and critical data science and machine learning projects across multiple verticals. He is currently associated with Amazon Web Services as a research scientist. He was born and raised in Pune, India and graduated from the University of Pune with a major in Information Technology Engineering. He started his career with Mu Sigma Inc., the world's largest pure-play analytics provider and worked with the leaders of many Fortune 50 clients. He later worked with Flutura - an IoT analytics startup and GE. He currently resides in Vancouver, BC. Apart from writing books on decision science and IoT, Jojo has also been a technical reviewer for various books on machine learning, deep learning and business analytics with Apress and Packt publications. He is an active data science tutor and maintains a blog at http: //blog.jojomoolayil.com.