The Keras package for R brings the power of deep learning to R users. Deep Learning with R in Motion locks in the essentials of deep learning and teaches you the techniques you'll need to start building and using your own neural …
The Keras package for R brings the power of deep learning to R users. Deep Learning with R in Motion locks in the essentials of deep learning and teaches you the techniques you'll need to start building and using your own neural networks for text and image processing.
Instructor Rick Scavetta takes you through a hands-on ride through the powerful Keras package, a TensorFlow API. You'll start by digging into case studies for how and where to use deep learning. Then, you'll master the essential components of a deep learning neural network as you work hands-on through your first examples. You'll continue by exploring dense and recurrent neural networks, convolutional and generative networks, and how they all work together.
And that's just the beginning! You'll go steadily deeper, making your network more robust and efficient. As your work through each module, you'll train your network and pick up the best practices used by experts like expert instructor Rick Scavetta, Keras library creator and author of Deep Learning in Python
François Chollet, and JJ Allaire, founder of RStudio, creator of the R bindings for Keras, and coauthor of Deep Learning in R! You'll beef up your skills as you practice with R-based applications in computer vision, natural-language processing, and generative models, ready for the real-world.
Machine learning has made remarkable progress in recent years. Deep learning systems have revolutionized image recognition, natural-language processing, and other applications for identifying complex patterns in data. The Keras library provides data scientists and developers working in R a state-of-the-art toolset for tackling deep learning tasks!
Inside:
GETTING STARTED
Welcome to the Video Series 00:07:18
What is Deep Learning? 00:06:15
The Landscape of Deep Learning 00:02:27
The Landscape of Machine Learning 00:05:09
The Two Golden Hypotheses 00:04:44
The 4 Types of Machine Learning 00:05:24
MNIST CASE STUDY
Unit Introduction 00:02:28
The MNIST dataset 00:10:44
A first look at a neural network 00:06:37
The 4 steps of Deep Learning, part 1 00:06:50
The 4 steps of Deep Learning, part 2 00:06:21
The Uses of Derivatives 00:05:12
From Derivatives to Gradients 00:04:32
Momentum in Mini-batch Stochastic Gradient Descent 00:05:52
The 4 steps of Deep Learning, part 3 00:03:26
Basic Model Evaluation 00:03:43
THREE CASE STUDIES FOR DEEP LEARNING
Unit Introduction 00:02:09
The story so far 00:02:48
The Reuters Newswire dataset: data preparation 00:06:32
The Reuters Newswire dataset: model definition and evaluation 00:07:37
The Reuters Newswire dataset: reanalysis 00:07:05
The IMDB Dataset: Data preparation, model definition, and evaluation 00:06:24
The IMDB Dataset: reanalysis 00:05:59
The Boston Housing Dataset: data preparation and model definition 00:06:43
The Boston Housing Dataset: K-fold cross validation and evaluation 00:05:18
Summary of the case studies 00:02:33
MODEL EVALUATION AND THE UNIVERSAL WORKFLOW
Review of the landscape 00:02:26
Validation: 3 varieties 00:05:35
Model Evaluation 00:07:54
Data Pre-processing 00:04:35
The machine learning universal workflow and Part 1 wrap-up 00:08:15
COMPUTER VISION
Unit Intro 00:01:16
Intro to Computer Vision 00:03:04
Convnets on MNIST 00:09:19
Convnets 1: Define Convnets from Scratch 00:05:42
Convnets 1: Import, Compile, and Train 00:02:58
Convnets 2: Data Augmentation 00:04:18
Convnets 3: Pre-Trained Intro 00:04:20
Convnets 3: Pre-Trained Code 00:03:52
TEXT AND SENTENCES
Introduction to Text and Sequences 00:06:33
Word Embeddings from Scratch 00:02:00
Pre-Trained Word Embeddings 00:05:24
RNNs on the IMDb Dataset 00:03:02
LSTMs on the IMDb Dataset 00:02:30
BEST PRACTICES & CONCLUSION ON PATTERN MATCHING
Chapter Intro 00:00:59
Idiosyncratic Structures 00:03:19
Callbacks and TensorBoard 00:01:01
A Review of Best Practices 00:03:30