This book is meant mainly for two types of readers: professionals who use R in their daily work and want to know some of the tricks and best practices that we have discovered in our projects, and anyone who wants to jump into the field of deep learning with concrete use cases in mind. This is not a tutorial in R or machine learning; it is rather a real-world showcase of how deep learning can be used in industry, with examples borrowed from our own experience.

Chapter 1, Handwritten Digit Recognition Using Convolutional Neural Networks, is where we start working on the first project of the book. We will begin with a recap of logistic regression and multilayer perceptron. We'll solve the problem with these two algorithms. We will then move on to the biologically inspired variants of multilayer perceptron—convolutional neural networks (CNNs). We will also cover the basics and core concepts of deep learning.

Chapter 2, Traffic Sign Recognition for Intelligent Vehicles, explains how to use CNNs for another application—traffic sign detection. We will also cover several important concepts of deep learning in this chapter and get readers familiar with other popular frameworks and libraries, such as Keras and TensorFlow. We will also introduce the dropout technique as a regularization approach and utilize data augmentation techniques to deal with a lack of training data.

Chapter 3, Fraud Detection with Autoencoders, introduces a type of deep learning model that can be used for anomaly detection. Outliers can be found within a collection of images, a text corpus, or transactional data. We will dive into applications of autoencoders and how they can be used for outlier detection in this domain. 

Chapter 4, Text Generation Using Recurrent Neural Networks, introduces different models of neural networks that try to capture the elusive properties of memory and abstraction to produce powerful models. We will apply different methods to tackle the text generation problem and suggest directions of further exploration. 

Chapter 5, Sentiment Analysis with Word Embeddings, shows how to use the popular GloVe algorithm for sentiment analysis, as well as other, less abstract tools. Although this algorithm is, strictly speaking, not a deep learning application, it belongs to the modern (as of 2018) toolkit of the data scientist, and it can be combined with other deep learning techniques.

You should be comfortable with R and RStudio and have some knowledge of college-level mathematics (calculus and linear algebra). Working knowledge of basic machine learning algorithms for classification, regression problems, and clustering might be helpful, but it is not strictly required.

> for (i in 1:16) {
+ outputData <- as.array
(executor$ref.outputs$activation15_output)[,,i,1]
+ image(outputData, xaxt='n', yaxt='n',
col=grey.colors(255)
+ )
+ }

library(R6)
logit <- function(x){
 1/(1+exp(-x))
}

Feedback from our readers is always welcome.

General feedback: Email [email protected] and mention the book title in the subject of your message. If you have questions about any aspect of this book, please email us at [email protected].

Errata: Although we have taken every care to ensure the accuracy of our content, mistakes do happen. If you have found a mistake in this book, we would be grateful if you would report this to us. Please visit www.packtpub.com/submit-errata, selecting your book, clicking on the Errata Submission Form link, and entering the details.

Piracy: If you come across any illegal copies of our works in any form on the Internet, we would be grateful if you would provide us with the location address or website name. Please contact us at [email protected] with a link to the material.

If you are interested in becoming an author: If there is a topic that you have expertise in and you are interested in either writing or contributing to a book, please visit authors.packtpub.com.