What this book covers

Chapter 1, Why Deep Learning?, provides an overview of deep learning. We begin with the history of deep learning, its rise, and its recent advances in certain fields. We will also describe some of its challenges, as well as its future potential.

Chapter 2, Getting Yourself Ready for Deep Learning, is a starting point to set oneself up for experimenting with and applying deep learning techniques in the real world. We will answer the key questions as to what skills and concepts are needed to get started with deep learning. We will cover some basic concepts of linear algebra, the hardware requirements for deep learning implementation, as well as some of its popular software frameworks. We will also take a look at setting up a deep learning system from scratch on a cloud-based GPU instance.

Chapter 3, Getting Started with Neural Networks, focuses on the basics of neural networks, including input/output layers, hidden layers, and how networks learn through forward and backpropagation. We will start with the standard multilayer perceptron networks and their building blocks, and illustrate how they learn step by step. We will also introduce a few popular standard models, such as Convolutional Neural Networks (CNNs), Restricted Boltzmann Machines (RBM), recurrent neural networks (RNNs), as well as a variation of them is called Long Short-Term Memory (LSTM).

Chapter 4, Deep Learning in Computer Vision, explains CNNs in more detail. We will go over the core concepts that are essential to the workings of CNNs and how they can be used to solve real-world computer vision problems. We will also look at some of the popular CNN architectures and implement a basic CNN using TensorFlow. 

Chapter 5, NLP – Vector Representation, covers the basics of NLP for deep learning. This chapter will describe the popular word embedding techniques used for feature representation in NLP. It will also cover popular models such as Word2Vec, Glove, and FastText. This chapter also includes an example of embedding training using TensorFlow.

Chapter 6, Advanced Natural Language Processing, takes a more model-centric approach to text processing. We will go over some of the core models, such as RNNs and LSTM networks. We will implement a sample LSTM using TensorFlow and describe the foundational architecture behind commonly used text processing applications of LSTM.

Chapter 7, Multimodality, introduces some fundamental progress in the multimodality of using deep learning. This chapter also shares some novel, advanced multimodal applications of deep learning.

Chapter 8, Deep Reinforcement Learning, covers the basics of reinforcement learning. It illustrates how deep learning can be applied to improve reinforcement learning in general. This chapter goes through the basic implementation of a basic deep reinforcement learning using TensorFlow and will also discuss some of its popular applications.

Chapter 9, Deep Learning Hacks, empowers readers by providing many practical tips that can be applied when using deep learning, such as the best practices for network weight initialization, learning parameter tuning, how to prevent overfitting, and how to prepare your data for better learning when facing data challenges.

Chapter 10, Deep Learning Trends, summarizes some of the upcoming ideas in deep learning. It looks at some of the upcoming trends in newly developed algorithms, as well as some of the new applications of deep learning.