Neural Network Programming with TensorFlow电子书

Title Page

Copyright

Neural Network Programming with TensorFlow

Credits

About the Authors

About the Reviewer

www.PacktPub.com

Why subscribe?

Customer Feedback

Preface

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

Maths for Neural Networks

Understanding linear algebra

Environment setup

Setting up the Python environment in Pycharm

Linear algebra structures

Scalars, vectors, and matrices

Tensors

Operations

Vectors

Matrices

Matrix multiplication

Trace operator

Matrix transpose

Matrix diagonals

Identity matrix

Inverse matrix

Solving linear equations

Singular value decomposition

Eigenvalue decomposition

Principal Component Analysis

Calculus

Gradient

Hessian

Determinant

Optimization

Optimizers

Summary

Deep Feedforward Networks

Defining feedforward networks

Understanding backpropagation

Implementing feedforward networks with TensorFlow

Analyzing the Iris dataset

Code execution

Implementing feedforward networks with images

Analyzing the effect of activation functions on the feedforward networks accuracy

Summary

Optimization for Neural Networks

What is optimization?

Types of optimizers

Gradient descent

Different variants of gradient descent

Algorithms to optimize gradient descent

Which optimizer to choose

Optimization with an example

Summary

Convolutional Neural Networks

An overview and the intuition of CNN

Single Conv Layer Computation

CNN in TensorFlow

Image loading in TensorFlow

Convolution operations

Convolution on an image

Strides

Pooling

Max pool

Example code

Average pool

Image classification with convolutional networks

Defining a tensor for input images and the first convolution layer

Input tensor

First convolution layer

Second convolution layer

Third convolution layer

Flatten the layer

Fully connected layers

Defining cost and optimizer

Optimizer

First epoch

Plotting filters and their effects on an image

Summary

Recurrent Neural Networks

Introduction to RNNs

RNN implementation

Computational graph

RNN implementation with TensorFlow

Computational graph

Introduction to long short term memory networks

Life cycle of LSTM

LSTM implementation

Computational graph

Sentiment analysis

Word embeddings

Sentiment analysis with an RNN

Computational graph

Summary

Generative Models

Generative models

Discriminative versus generative models

Types of generative models

Autoencoders

GAN

Sequence models

GANs

GAN with an example

Types of GANs

Vanilla GAN

Conditional GAN

Info GAN

Wasserstein GAN

Coupled GAN

Summary

Deep Belief Networking

Understanding deep belief networks

DBN implementation

Class initialization

RBM class

Pretraining the DBN

Model training

Predicting the label

Finding the accuracy of the model

DBN implementation for the MNIST dataset

Loading the dataset

Input parameters for a DBN with 256-Neuron RBM layers

Output for a DBN with 256-neuron RBN layers

Effect of the number of neurons in an RBM layer in a DBN

An RBM layer with 512 neurons

An RBM layer with 128 neurons

Comparing the accuracy metrics

DBNs with two RBM layers

Classifying the NotMNIST dataset with a DBN

Summary

Autoencoders

Autoencoder algorithms

Under-complete autoencoders

Dataset

Basic autoencoders

Autoencoder initialization

AutoEncoder class

Basic autoencoders with MNIST data

Basic autoencoder plot of weights

Basic autoencoder recreated images plot

Basic autoencoder full code listing

Basic autoencoder summary

Additive Gaussian Noise autoencoder

Autoencoder class

Additive Gaussian Autoencoder with the MNIST dataset

Training the model

Plotting the weights

Plotting the reconstructed images

Additive Gaussian autoencoder full code listing

Comparing basic encoder costs with the Additive Gaussian Noise autoencoder

Additive Gaussian Noise autoencoder summary

Sparse autoencoder

KL divergence

KL divergence in TensorFlow

Cost of a sparse autoencoder based on KL Divergence

Complete code listing of the sparse autoencoder

Sparse autoencoder on MNIST data

Comparing the Sparse encoder with the Additive Gaussian Noise encoder

Summary

Research in Neural Networks

Avoiding overfitting in neural networks

Problem statement

Solution

Results

Large-scale video processing with neural networks

Resolution improvements

Feature histogram baselines

Quantitative results

Named entity recognition using a twisted neural network

Example of a named entity recognition

Defining Twinet

Results

Bidirectional RNNs

BRNN on TIMIT dataset

Summary

Getting started with TensorFlow

Environment setup

TensorFlow comparison with Numpy

Computational graph

Graph

Session objects

Variables

Scope

Data input

Placeholders and feed dictionaries

Auto differentiation

TensorBoard