Deep Learning with R. First Edition (MEAP). By François Chollet. Published by Manning.
I bought this book at the end of 2017. It was an early released from Manning Publications. And I have to say that every cent I did spend on this book was worthwhile. In fact, I spent most of my Christmas time enjoying and dissecting it.
Let’s break it into the chapters:
- Chapter 1: What is deep learning?
- Chapter 2: Before we begin: the mathematical blocks of neural networks
- Chapter 3: Getting started with Neural networks
- Chapter 4: Fundamentals of machine learning
- Chapter 5: Deep learning for computer vision
- Chapter 6: Deep learning for text and sequences
- Chapter 7: Advanced deep-learning best practices
- Chapter 8: Generative deep learning
- Chapter 9: Conclusions
The structure of the book is perfect for newbies in the deep learning field. It goes from more straightforward or most known algorithms and methodologies to the most recent and high-level ones.
This book has both, an R and a Python version. I bought the R version as I was quite curious about doing a DL project with R (which is not the most common language for it). I found this little jewel while I was trying to code an LSTM for deep writing on Cervante’s style. I did it as I could reuse a Python script from the deep learning nano degree at Udacity, but I did want to go forward and replicate the script from scratch in R. I am an R lady , and I was hoping I could make an introduction to deep learning with R someday.
The book contains a clean, step-by-step coded examples that are available in a Github repository; the reason why I think this is pure gold is the theory that goes along the cases. I was not new at all in the field of deep learning, and I also learned something in this reading.
LSTM models, recurrent networks, GANs, transfer learning and how to tune the parameters and hyperparameters were concepts that I heard about before. But the first time I learned about bidirectional sequence models was in this book.