Book Image

Hands-On Deep Learning with R

By : Michael Pawlus, Rodger Devine

Book Image

Hands-On Deep Learning with R

By: Michael Pawlus, Rodger Devine

Overview of this book

Deep learning enables efficient and accurate learning from a massive amount of data. This book will help you overcome a number of challenges using various deep learning algorithms and architectures with R programming. This book starts with a brief overview of machine learning and deep learning and how to build your first neural network. You’ll understand the architecture of various deep learning algorithms and their applicable fields, learn how to build deep learning models, optimize hyperparameters, and evaluate model performance. Various deep learning applications in image processing, natural language processing (NLP), recommendation systems, and predictive analytics will also be covered. Later chapters will show you how to tackle recognition problems such as image recognition and signal detection, programmatically summarize documents, conduct topic modeling, and forecast stock market prices. Toward the end of the book, you will learn the common applications of GANs and how to build a face generation model using them. Finally, you’ll get to grips with using reinforcement learning and deep reinforcement learning to solve various real-world problems. By the end of this deep learning book, you will be able to build and deploy your own deep learning applications using appropriate frameworks and algorithms.

Preface

Who this book is for

What this book covers

To get the most out of this book

Section 1: Deep Learning Basics

Section 1: Deep Learning Basics

Free Chapter

Machine Learning Basics

Machine Learning Basics

An overview of machine learning

Preparing data for modeling

Training a model on prepared data

Evaluating model results

Improving model results

Reviewing different algorithms

Setting Up R for Deep Learning

Setting Up R for Deep Learning

Technical requirements

Installing the packages

Preparing a sample dataset

Exploring Keras

Exploring MXNet

Exploring ReinforcementLearning and RBM

Comparing the deep learning libraries

Artificial Neural Networks

Artificial Neural Networks

Technical requirements

Contrasting deep learning with machine learning

Comparing neural networks and the human brain

Utilizing bias and activation functions within hidden layers

Surveying activation functions

Creating a feedforward network

Augmenting our neural network with backpropagation

Section 2: Deep Learning Applications

Section 2: Deep Learning Applications

CNNs for Image Recognition

CNNs for Image Recognition

Technical requirements

Image recognition with shallow nets

Image recognition with convolutional neural networks

Enhancing the model with additional layers

Choosing the most appropriate activation function

Selecting optimal epochs using dropout and early stopping

Multilayer Perceptron for Signal Detection

Multilayer Perceptron for Signal Detection

Technical requirements

Understanding multilayer perceptrons

Preparing and preprocessing data

Deciding on the hidden layers and neurons

Training and evaluating the model

Neural Collaborative Filtering Using Embeddings

Neural Collaborative Filtering Using Embeddings

Technical requirements

Introducing recommender systems

Collaborative filtering with neural networks

Exploring embeddings

Preparing, preprocessing, and exploring data

Performing exploratory data analysis

Creating user and item embeddings

Building and training a neural recommender system

Evaluating results and tuning hyperparameters

Deep Learning for Natural Language Processing

Deep Learning for Natural Language Processing

Formatting data using tokenization

Cleaning text to remove noise

Applying word embeddings to increase usable data

Clustering data into topic groups

Summarizing documents using model results

Creating an RBM

Defining the Gibbs sampling rate

Speeding up sampling with contrastive divergence

Computing free energy for model evaluation

Stacking RBMs to create a deep belief network

Long Short-Term Memory Networks for Stock Forecasting

Long Short-Term Memory Networks for Stock Forecasting

Technical requirements

Understanding common methods for stock market prediction

Preparing and preprocessing data

Configuring a data generator

Training and evaluating the model

Tuning hyperparameters to improve performance

Generative Adversarial Networks for Faces

Generative Adversarial Networks for Faces

Technical requirements

An overview of GANs

Defining the generator model

Defining the discriminator model

Preparing and preprocessing a dataset

Training and evaluating the model

Section 3: Reinforcement Learning

Section 3: Reinforcement Learning

Reinforcement Learning for Gaming

Reinforcement Learning for Gaming

Technical requirements

Understanding the concept of reinforcement learning

Preparing and processing data

Configuring the reinforcement agent

Tuning hyperparameters

Deep Q-Learning for Maze Solving

Deep Q-Learning for Maze Solving

Technical requirements

Creating an environment for reinforcement learning

Defining an agent to perform actions

Building a deep Q-learning model

Running the experiment

Improving performance with policy functions

Other Books You May Enjoy

Other Books You May Enjoy

Leave a review - let other readers know what you think

Customer Reviews

5 star

0

4 star

0

3 star

0

2 star

0

1 star

0

Defining the Gibbs sampling rate

Gibbs sampling plays a key role in constructing an RBM, so we will take a moment here to define this sampling type. We will briefly walk through a couple of quick concepts that lead to how to perform Gibbs sampling and why it matters for this type of modeling. With RBM models, we are first using a neural network to map our input or visible units to hidden units, which can be thought of as latent features. After training our model, we want to either take a new visible unit and define the probability that it belongs to the hidden units in the model, or do the reverse. We also want this to be computationally efficient, so we use a Monte Carlo approach.

Monte Carlo methods involve sampling random points to approximate an area or distribution. A classic example involves drawing a 10-by-10 inch square and inside this square draw a circle. We know...