book

Deep Learning with TensorFlow - Second Edition

Name: Deep Learning with TensorFlow - Second Edition
ISBN: 9781788831109

by Giancarlo Zaccone, Vihan Jain, Md. Rezaul Karim, Motaz Saad

March 2018

Intermediate to advanced

484 pages

10h 31m

English

Packt Publishing

Read now

Unlock full access

Deep Learning with TensorFlow - Second Edition
Table of Contents
Deep Learning with TensorFlow - Second Edition
Why subscribe?
PacktPub.com
Contributors
About the authors
About the reviewers
Packt is Searching for Authors Like You
Preface
Who this book is for
What this book covers
To get the most out of this book
Download the example code filesDownload the color imagesConventions used

Get in touch
Reviews
1. Getting Started with Deep Learning
A soft introduction to machine learningSupervised learningUnbalanced dataUnsupervised learningReinforcement learningWhat is deep learning?
Artificial neural networks
The biological neuronsThe artificial neuron
How does an ANN learn?
ANNs and the backpropagation algorithmWeight optimizationStochastic gradient descent
Neural network architectures
Deep Neural Networks (DNNs)Multilayer perceptronDeep Belief Networks (DBNs)Convolutional Neural Networks (CNNs)AutoEncodersRecurrent Neural Networks (RNNs)Emergent architectures
Deep learning frameworks
Summary
2. A First Look at TensorFlow
A general overview of TensorFlow
What's new from TensorFlow v1.6 forwards?
Nvidia GPU support optimizedIntroducing TensorFlow LiteEager executionOptimized Accelerated Linear Algebra (XLA)
Installing and configuring TensorFlow
TensorFlow computational graph
TensorFlow code structure
Eager execution with TensorFlow
Data model in TensorFlow
TensorRank and shapeData typeVariablesFetchesFeeds and placeholders
Visualizing computations through TensorBoard
How does TensorBoard work?
Linear regression and beyond
Linear regression revisited for a real dataset
Summary
3. Feed-Forward Neural Networks with TensorFlow
Feed-forward neural networks (FFNNs)Feed-forward and backpropagationWeights and biasesActivation functionsUsing sigmoidUsing tanhUsing ReLUUsing softmax
Implementing a feed-forward neural network
Exploring the MNIST datasetSoftmax classifier
Implementing a multilayer perceptron (MLP)
Training an MLPUsing MLPsDataset descriptionPreprocessingA TensorFlow implementation of MLP for client-subscription assessmentDeep Belief Networks (DBNs)Restricted Boltzmann Machines (RBMs)Construction of a simple DBNUnsupervised pre-trainingSupervised fine-tuningImplementing a DBN with TensorFlow for client-subscription assessment
Tuning hyperparameters and advanced FFNNs
Tuning FFNN hyperparametersNumber of hidden layersNumber of neurons per hidden layerWeight and biases initializationSelecting the most suitable optimizerGridSearch and randomized search for hyperparameters tuningRegularizationDropout optimization
Summary
4. Convolutional Neural Networks
Main concepts of CNNs
CNNs in action
LeNet5
Implementing a LeNet-5 step by step
AlexNetTransfer learningPretrained AlexNet
Dataset preparation
Fine-tuning implementation
VGGArtistic style learning with VGG-19Input imagesContent extractor and lossStyle extractor and lossMerger and total lossTraining
Inception-v3
Exploring Inception with TensorFlow
Emotion recognition with CNNs
Testing the model on your own imageSource code
Summary
5. Optimizing TensorFlow Autoencoders
How does an autoencoder work?
Implementing autoencoders with TensorFlow
Improving autoencoder robustness
Implementing a denoising autoencoderImplementing a convolutional autoencoderEncoderDecoder
Fraud analytics with autoencoders
Description of the datasetProblem descriptionExploratory data analysisTraining, validation, and testing set preparationNormalizationAutoencoder as an unsupervised feature learning algorithmEvaluating the model
Summary
6. Recurrent Neural Networks
Working principles of RNNsImplementing basic RNNs in TensorFlowRNN and the long-term dependency problemBi-directional RNNs
RNN and the gradient vanishing-exploding problem
LSTM networksGRU cell
Implementing an RNN for spam prediction
Data description and preprocessing
Developing a predictive model for time series data
Description of the datasetPre-processing and exploratory analysisLSTM predictive modelModel evaluation
An LSTM predictive model for sentiment analysis
Network designLSTM model trainingVisualizing through TensorBoardLSTM model evaluation
Human activity recognition using LSTM model
Dataset descriptionWorkflow of the LSTM model for HARImplementing an LSTM model for HAR
Summary
7. Heterogeneous and Distributed Computing
GPGPU computingThe GPGPU historyThe CUDA architectureThe GPU programming model
The TensorFlow GPU setup
Update TensorFlowGPU representationUsing a GPUGPU memory managementAssigning a single GPU on a multi-GPU systemThe source code for GPU with soft placementUsing multiple GPUs
Distributed computing
Model parallelismData parallelism
The distributed TensorFlow setup
Summary
8. Advanced TensorFlow Programming
tf.estimatorEstimatorsGraph actionsParsing resourcesFlower predictions
TFLearn
InstallationTitanic survival predictor
PrettyTensor
Chaining layersNormal modeSequential modeBranch and joinDigit classifier
Keras
Keras programming modelsSequential modelSentiment classification of movie reviewsFunctional APISqueezeNet
Summary
9. Recommendation Systems Using Factorization Machines
Recommendation systemsCollaborative filtering approachesContent-based filtering approachesHybrid recommender systemsModel-based collaborative filtering
Movie recommendation using collaborative filtering
The utility matrixDescription of the datasetRatings dataMovies dataUsers dataExploratory analysis of the MovieLens datasetImplementing a movie RETraining the model with the available ratingsInferencing the saved modelGenerating the user-item tableClustering similar moviesMovie rating prediction by usersFinding top k moviesPredicting top k similar moviesComputing user-user similarityEvaluating the recommender system
Factorization machines for recommendation systems
Factorization machinesCold-start problem and collaborative-filtering approachesProblem definition and formulationDataset descriptionWorkflow of the implementationPreprocessingTraining the FM model
Improved factorization machines
Neural factorization machinesDataset descriptionUsing NFM for the movie recommendationModel trainingModel evaluation
Summary
10. Reinforcement Learning
The RL problem
OpenAI Gym
OpenAI environmentsThe env classInstalling and running OpenAI Gym
The Q-Learning algorithm
The FrozenLake environment
Deep Q-learning
Deep Q neural networksThe Cart-Pole problemDeep Q-Network for the Cart-Pole problemThe Experience Replay methodExploitation and explorationThe Deep Q-Learning training algorithm
Summary
Other Books You May Enjoy
Leave a review – let other readers know what you think
Index

Content preview from Deep Learning with TensorFlow - Second Edition

RNN and the gradient vanishing-exploding problem

Gradients for deeper layers are calculated as products of many gradients of activation functions in the multi-layer network. When those gradients are small or zero, it will easily vanish. On the other hand, when they are bigger than 1, it will possibly explode. So, it becomes very hard to calculate and update.

Let's explain them in more detail:

If the weights are small, it can lead to a situation called vanishing gradients, where the gradient signal gets so small that learning either becomes very slow or stops working altogether. This is often referred to as vanishing gradients.
If the weights in this matrix are large, it can lead to a situation where the gradient signal is so large that it can cause ...

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.

Read now

Unlock full access

More than 5,000 organizations count on O’Reilly

O’Reilly covers everything we've got, with content to help us build a world-class technology community, upgrade the capabilities and competencies of our teams, and improve overall team performance as well as their engagement.

Julian F.

Head of Cybersecurity

I wanted to learn C and C++, but it didn't click for me until I picked up an O'Reilly book. When I went on the O’Reilly platform, I was astonished to find all the books there, plus live events and sandboxes so you could play around with the technology.

Addison B.

Field Engineer

I’ve been on the O’Reilly platform for more than eight years. I use a couple of learning platforms, but I'm on O'Reilly more than anybody else. When you're there, you start learning. I'm never disappointed.

Amir M.

Data Platform Tech Lead

I'm always learning. So when I got on to O'Reilly, I was like a kid in a candy store. There are playlists. There are answers. There's on-demand training. It's worth its weight in gold, in terms of what it allows me to do.

Mark W.

Embedded Software Engineer

Deep Learning with TensorFlow 2 and Keras - Second Edition

Publisher Resources

ISBN: 9781788831109

Cloud Computing

Data Engineering

Data Science

AI & ML

Programming Languages

Software Architecture

IT/Ops

Security

Design

Business

Soft Skills

Deep Learning with TensorFlow - Second Edition

by Giancarlo Zaccone, Vihan Jain, Md. Rezaul Karim, Motaz Saad

RNN and the gradient vanishing-exploding problem

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.