Deep Learning with Keras电子书

Title Page

Credits

About the Authors

About the Reviewer

www.PacktPub.com

Customer Feedback

Preface

Mission

How deep learning is different from machine learning and artificial intelligence

What this book covers

What you need for this book

Who this book is for

Conventions

Reader feedback

Customer support

Downloading the example code

Downloading the color images of this book

Errata

Piracy

Questions

Neural Networks Foundations

Perceptron

The first example of Keras code

Multilayer perceptron - the first example of a network

Problems in training the perceptron and a solution

Activation function - sigmoid

Activation function - ReLU

Activation functions

A real example - recognizing handwritten digits

One-hot encoding - OHE

Defining a simple neural net in Keras

Running a simple Keras net and establishing a baseline

Improving the simple net in Keras with hidden layers

Further improving the simple net in Keras with dropout

Testing different optimizers in Keras

Increasing the number of epochs

Controlling the optimizer learning rate

Increasing the number of internal hidden neurons

Increasing the size of batch computation

Summarizing the experiments run for recognizing handwritten charts

Adopting regularization for avoiding overfitting

Hyperparameters tuning

Predicting output

A practical overview of backpropagation

Towards a deep learning approach

Summary

Keras Installation and API

Installing Keras

Step 1 - install some useful dependencies

Step 2 - install Theano

Step 3 - install TensorFlow

Step 4 - install Keras

Step 5 - testing Theano, TensorFlow, and Keras

Configuring Keras

Installing Keras on Docker

Installing Keras on Google Cloud ML

Installing Keras on Amazon AWS

Installing Keras on Microsoft Azure

Keras API

Getting started with Keras architecture

What is a tensor?

Composing models in Keras

Sequential composition

Functional composition

An overview of predefined neural network layers

Regular dense

Recurrent neural networks - simple, LSTM, and GRU

Convolutional and pooling layers

Regularization

Batch normalization

An overview of predefined activation functions

An overview of losses functions

An overview of metrics

An overview of optimizers

Some useful operations

Saving and loading the weights and the architecture of a model

Callbacks for customizing the training process

Checkpointing

Using TensorBoard and Keras

Using Quiver and Keras

Summary

Deep Learning with ConvNets

Deep convolutional neural network - DCNN

Local receptive fields

Shared weights and bias

Pooling layers

Max-pooling

Average pooling

ConvNets summary

An example of DCNN - LeNet

LeNet code in Keras

Understanding the power of deep learning

Recognizing CIFAR-10 images with deep learning

Improving the CIFAR-10 performance with deeper a network

Improving the CIFAR-10 performance with data augmentation

Predicting with CIFAR-10

Very deep convolutional networks for large-scale image recognition

Recognizing cats with a VGG-16 net

Utilizing Keras built-in VGG-16 net module

Recycling pre-built deep learning models for extracting features

Very deep inception-v3 net used for transfer learning

Summary

Generative Adversarial Networks and WaveNet

What is a GAN?

Some GAN applications

Deep convolutional generative adversarial networks

Keras adversarial GANs for forging MNIST

Keras adversarial GANs for forging CIFAR

WaveNet - a generative model for learning how to produce audio

Summary

Word Embeddings

Distributed representations

word2vec

The skip-gram word2vec model

The CBOW word2vec model

Extracting word2vec embeddings from the model

Using third-party implementations of word2vec

Exploring GloVe

Using pre-trained embeddings

Learn embeddings from scratch

Fine-tuning learned embeddings from word2vec

Fine-tune learned embeddings from GloVe

Look up embeddings

Summary

Recurrent Neural Network — RNN

SimpleRNN cells

SimpleRNN with Keras - generating text

RNN topologies

Vanishing and exploding gradients

Long short term memory - LSTM

LSTM with Keras - sentiment analysis

Gated recurrent unit - GRU

GRU with Keras - POS tagging

Bidirectional RNNs

Stateful RNNs

Stateful LSTM with Keras - predicting electricity consumption

Other RNN variants

Summary

Additional Deep Learning Models

Keras functional API

Regression networks

Keras regression example - predicting benzene levels in the air

Unsupervised learning - autoencoders

Keras autoencoder example - sentence vectors

Composing deep networks

Keras example - memory network for question answering

Customizing Keras

Keras example - using the lambda layer

Keras example - building a custom normalization layer

Generative models

Keras example - deep dreaming

Keras example - style transfer

Summary

AI Game Playing

Reinforcement learning

Maximizing future rewards

Q-learning

The deep Q-network as a Q-function

Balancing exploration with exploitation

Experience replay, or the value of experience

Example - Keras deep Q-network for catch

The road ahead

Summary

Conclusion

Keras 2.0 - what is new

Installing Keras 2.0

API changes