Index – Deep Learning with TensorFlow 2 and Keras – Second Edition

Index

A

Accelerated Linear Algebra (XLA) 82

achievements, RL

AI controlled sailplanes 416

AlphaGo Zero 415

locomotion behavior 416

Action-Value function 411

activation functions

about 549

derivative of ReLU 551

derivative of sigmoid 549, 550

derivative of tanh 550

additive attention 330

adversarial training

about 192

Agent 408

AI controlled sailplanes 416

AlexNet 160

AlphaGo Zero 416

Amazon EC2

virtual machines, creating 448, 449

Amazon Machine Image (AMI)

reference link 449

selecting 449

Amazon Web Service (AWS)

about 439, 442, 443

reference link 442

services, reference link 443

Android Studio

installation link 467

reference link 466

App Engine

reference link 446

applications, GAN

album covers 217

bedroom 217

Application-Specific Integrated Circuit (ASIC)

about 573

Arduino Nano 33 BLE Sense

reference link 464

area under the cover receiver operating characteristic (AUC ROC) 507

Area Under the Curve (AUC) 507

Arm Cortex-M Series Processors

reference link 464

Artificial General Intelligence (AGI) 407, 491

artificial neural networks (ANNs) 5

AtrousConvolution 181

Attention mechanism

about 329, 330, 331

integrating, with seq2seq network 331, 332, 333, 334, 335, 336, 337

machine translation example 331, 332, 333, 334, 335, 336, 337

autoencoders

about 345, 346, 347

denoising autoencoders 358

sparse autoencoder 356, 357

stacked autoencoder 362

vanilla autoencoder 347

Auto-Encoding Variational Bayes

reference link 402

AutoGraph 61, 62

AutoKeras

about 497

automatic differentiation 569

AutoML 47

about 491, 492

achieving 492

benefits 492

AutoML Natural Language

AutoML Entity Extraction 498

AutoML Sentiment Analysis 498

AutoML Text Classification 498

Cloud Natural Language API 498

AutoML pipeline

data preparation 493

feature engineering 493

model generation 494, 495, 496

AutoML Translation

about 498

AutoML Video

AutoML Video Classification 498

AutoML Vision

about 498

AutoML Object Detection 498

average pooling 113

AWS services

AWS IoT 444

Elastic Beanstalk 444

Elastic Compute Cloud (EC2) 443

Lambda 444

SageMaker 444

B

backpropagation

about 551, 552, 553, 554

and convnets 566

and RNNs 566, 567, 568, 569

backstep 555, 556

forward step 554

limits 562

overview 48, 49

Backpropagation Through Time (BTT) 567

backstep

about 555, 556

neuron equation, from hidden layer to hidden layer 558, 559, 560, 561

neuron equation, from hidden layer to output layer 557, 558

backward pass 393

Bahdanau attention 330

batch computation

size, increasing 37

Batch Gradient Descent (BGD) 564, 565

BatchNormalization 40, 41

beginning-of-string (BOS) 318

Bidirectional Encoder Representations (BERT)

about 267

classifying 271, 272

fine-tuning 270

using, as feature extractor 269

using, as part of network 272, 273, 274, 276

bidirectional RNNs 289, 290

BiLingual Evaluation Understudy (BLEU) 325

Blender learning environment 417

Boston housing price dataset

reference link 97

C

Caffe

reference link 1

callbacks 67

Canned Estimators 94

capsule 187

Capsule Networks (CapsNets)

about 186

features 187, 188

verus CNN 187, 188

CartPole

DQN 424, 425, 426, 427, 428

reference link 424

Central Processing Units (CPUs)

about 571, 572

chain rule 548

character embeddings 260

CIFAR-10

used, for prediction 130

CIFAR-10 images

recognizing, with deep learning 122, 123, 124, 125

CIFAR-10 performance

improving, with data augmentation 128, 129, 130

improving, with deeper network 125, 126, 127, 128

CIFAR repository

reference link 122

classification head 140

classification task 101

CLEVR dataset

reference link 165

Cloud AutoML

reference link 446

Cloud Functions

reference link 446

Cloud IoT Core

reference link 446

cloud storage

creating, reference link 517

CNN architectures

about 160

AlexNet 160

DenseNets 161

HighwaysNets 161

residual networks 160

xception networks 161, 163

CNNs

issues 186

CNNs, composing for complex tasks

classification 140, 141

instance segmentation 145, 146, 147

localization 140, 141

object detection 142, 144, 145

semantic segmentation 141

Colaboratory Notebook

reference link 455

Colabs

TPUs, using with 580

color mapping

with SOM 387, 388, 389, 390, 391, 392

computational graph, TensorFlow 1.x

about 52

example 53, 54

execution 52

need for 52

program structure 51

Compute Engine

reference link 445

ConceptNet Numberbatch 247

config parameters, for convolution layer configuration

kernel size 185

padding 185

stride 185

constants 54

content-based attention 330

Contextualized Vectors (CoVe) 262

continuous backpropagation

history 543

Contrastive Divergence (CD) 392

ConvNet

about 184

convolution 110

convolutional autoencoder

about 362

implementing 362

used, for removing noise from images 362, 364, 365, 366

Convolutional Neural Network (CNN) 243, 478

about 184

using, for audio 179

using, for sentiment analysis 176, 177, 178, 179

convolutional neural networks (CNNs)

about 139

composing, for complex tasks 139

convolutional neural networks (ConvNets)

about 109

summarizing 113

TensorFlow 2.x 112

convolution operations

about 184

depthwise convolution 186

depthwise separable convolution 186

separable convolution 185

Corpus of Linguistic Acceptability (COLA) 271

cross entropy

about 562, 563, 564

derivative 562, 563, 564

Custom Estimators 94

custom layers, building

build() method, using 349

call() method, using 349

__init__() method, using 349

CycleGAN 210, 211, 212

in TensorFlow 2.0 218, 219, 220, 222, 223, 224, 225, 226, 227, 228

reference link 228

D

data cleansing 493

data preparation, AutoML

data cleansing 493

data synthesis 493

data synthesis 493

decision boundaries 101

Deep Averaging Network (DAN) 264

deep belief networks (DBNs) 397

deep convolutional networks

using, for large-scale image recognition 132, 133

Deep Convolutional Neural Network (DCNN)

about 110

EXAMPLE, LeNet 114

local receptive fields 110, 111

mathematical example 111, 112

pooling layers 113

shared weights and bias 111

deep convolution GAN (DCGAN)

about 198, 199

building, for generation of MNIST digits 200, 201, 204, 206, 207, 209

changes 198

Deep Deterministic Policy Gradient (DDPG)

about 436, 437

DeepDream network

creating 169, 170, 171, 172

Deep Inception-v3 Net

used, for transfer learning 151, 152, 153

DeepLab 146

deep learning

used, for recognizing CIFAR-10 images 122

deep learning approach 50

Deep Learning Containers

reference link 446

Deep Learning (DL) 478

history 543

deep learning (DL) model, on cloud

about 439

advantages 439, 440

AWS 442

categories 440

IBM cloud 447

Microsoft Azure 440

Platform as a Service (PaaS) 440

Software as a Service (SaaS) 440

DeepMind

reference link 180

Deep Q-Networks (DQNs)

about 422, 423, 424

for CartPole 424, 425, 426, 427, 428

used, for playing Atari game 429, 431, 432

variants 432

Deep Reinforcement Learning (DRL)

about 411

policy-based methods 412

value-based methods 411

working 412, 414, 415

denoising autoencoders

about 358

used, for clearing images 359

DenseNets 161

dependent variables

about 88

reference link 88

deprecated endpoints

reference link 80

derivatives 544

differentiation rules 548

Dilated Causal Convolutions 180

Dilated ConvNets 179, 181

dilated convolution

about 185

dilation rate 185

transposed convolution 185

distributed representations 233, 234

distributed training, TensorFlow 2.x

about 76

multiple GPUs 76, 77, 78

MultiWorkerMirroredStrategy 78

ParameterServerStrategy 78

TPUStrategy 78

dot product 232

double DQN 432

DQN variants

about 432

double DQN 432, 433

dueling DQN 433, 434, 435

rainbow 436

Dropout

used, for improving simple Net in TensorFlow 2.0 26, 27

dueling DQN 433, 434, 435

dynamic embeddings 261, 263

E

eager execution 60

edge computing

Federated Learning (FL) 474

edge TPU 578

Efficient Neural Architecture Search (ENAS) 495

eigen decomposition 375

Embedding Projector

about 379

Inspector Panel 380

Projections Panel 379

Embeddings from Language Models (ELMo) 262

end of sentence (EOS) 319

environments, OpenAI gym

algorithms 418

Atari 418

Box2D 418

classic control 418

MuJoCo 418

robotics 418

toy text 418

epochs

count, increasing 34

Epsilon-Greedy 414

Epsilon Greedy policy 426

estimator model 147

running, on GPUs 150, 151

Estimators 72, 73, 74

Evolutionary algorithm (EA) 496

Experience Replay method 414

exploration

versus exploitation 414

F

Facebook AI Research (FAIR) 260

False Positive Rate (FPR) 507

Fashion-MNIST

classifying, with tf.Keras 147, 148, 149, 150

reference link 147

Faster R-CNN

reference link 144

fastText

about 247

feature engineering, AutoML

about 493

feature construction 494

feature extraction 494

feature selection 493

feature extraction 494

feature selection 493

federated core (FC)

reference link 476

Federated Learning (FL)

reference link 477

Federated Learning (FL), edge computing

issues 475

overview 474, 475

Feed Forward Network (FFN) 339, 340

FigureQA dataset

reference link 165

first-generation TPU (TPU v1) 574, 575, 576

FlatBuffers

about 463

reference link 463

forward pass 392

G

GAN architectures

about 209

CycleGAN 210, 211, 212

InfoGAN 212, 213

SRGAN 209, 210

Gated Recurrent Unit (GRU) 279

Gazebo 417

GCP services

App Engine 446

Cloud AutoML 446

Cloud Functions 446

Cloud IoT Core 446

Compute Engine 445

Deep Learning Containers 446

Generative Adversarial Networks (GANs)

about 191, 192, 193

applications 214, 216, 217

building, with MNIST in TensorFlow 193, 194, 196, 197, 198

convergence, reference link 193

learning 193

reference link 191

Generative Pretraining (GPT) 267

Genetic Programming (GA) 496

gensim

installation link 240

reference link 240

used, for creating word embedding 239, 240

used, for exploring embedding space 240, 241, 242, 243

Global vectors for word representation (GloVe)

about 238, 247

download link 239

GloVe vectors 239

Google AutoML

integrating, with Kaggle 541, 542

Google Cloud AutoML

about 498

cost 540

reference link 498

Google Cloud AutoML Table

about 498

using 500, 501, 502, 503, 504, 506, 507, 509, 510, 513, 514

Google Cloud AutoML Text

using 525, 527, 529

Google Cloud AutoML Translation

using 529, 530, 531, 533

Google Cloud AutoML Video

using 534, 535, 536, 537, 539

Google Cloud AutoML Vision

using 514, 515, 516, 517, 518, 519, 520, 521, 522, 523

Google cloud console

reference link 587

Google cloud platform (GCP)

about 444

reference link 444

services 445

Google Cloud Platform (GCP) 517

Google Cloud Platform Pricing Calculator

reference link 500

Google Colab

CPUs 41, 42, 43

GPUs 41, 42, 43

playing with 41, 42, 43, 44

reference link 41

TPUs 41, 42, 43

Google Colaboratory

about 453, 454, 455

reference link 453

Google Compute Engine (GCE) 577

Google Kubernetes Engine (GKE) 577

Google Neural Machine Translation (Google NMT) 531

Google powered Dataset Search

reference link 82

GPUs

estimator model, running 150, 151

gradient descent 546

Gradient Descent (GD) 28

gradients 544, 545, 546

Gradle

reference link 469

Graphic Processing Units (GPUs)

about 571, 572

H

handwritten charts

experiments summary, for recognizing 38

handwritten digits

recognizing 15

reconstructing, with vanilla autoencoders 350, 352, 353, 354

hard clustering 384

hard update 415

hidden layer 392

HighwaysNets 161

hyperparameter tuning 47

about 494, 496

grid search 496

random search 496

I

IBM cloud

about 447

reference link 447

ImageNet ILSVRC-2012

reference link 132

images

clearing, denoising autoencoder used 359, 360, 361, 362

noise, removing with convolutional autoencoder 362, 364, 366, 367

independent variables

about 88

reference link 88

InfoGAN 212, 213

Information Retrieval (IR) 232

internal hidden neurons

count, increasing 36

Internet of Things (IoT) 578

IoT (Internet of Things) 461

Item2Vec 254

J

Java Caffe

reference link 132

Jupyter Notebooks, on cloud

about 451

Google Colaboratory 453, 454, 455

Microsoft Azure Notebook 456

SageMaker 453

Just In Time (JIT) compilation 62

K

Kaggle

Google AutoML, integrating into 541, 542

reference link 529

Kaggle VQA challenge

reference link 165

Keras 83

about 3

pooling layers, reference link 113

Keras APIs

about 63

Functional API 64, 66

model subclassing 66, 67

Sequential API 63

Keras applications

about 159

Keras Applications

reference link 152

k-means clustering

about 380

in TensorFlow 2.0 381, 382, 384

variations 384

working 380

Kohonen networks 384

Kullback-Leiber (KL) divergence 357

L

language model based embeddings

about 265, 266, 267, 268

BERT, fine-tuning 270

BERT, used for classification 271

BERT, using as feature extractor 268, 269

BERT, using as part of network 272, 273, 275, 276

Large Movie Review Dataset v1.0

reference link 472

Latent Semantic Analysis (LSA) 232

learning with a critic 408

left singular matrix 376

LeNet

about 114

deep learning 121, 122

in TensorFlow 2.0 114, 115, 116, 117, 118, 120

libraries, TFX

ML Metadata (MLMD) 459

TensorFlow 459

TensorFlow Data Validation (TFDV) 459

TensorFlow Metadata (TFMD) 459

TensorFlow Model Analysis (TFMA) 459

TensorFlow Transform (TFT) 459

linear regression

about 88

multiple linear regression 93

multivariate linear regression 93

simple linear regression 89, 90, 91, 92

used, for making prediction 88

used, for predicting house price 97, 98, 100, 101

logistic regression

using 102

using, on MNIST dataset 104, 105, 107

long short term memory (LSTM) 285, 286, 287

Long Short Term Memory (LSTMs) 262

loss functions

reference link 19

Luongs attention 330

M

Machine Learning (ML) models 478

Malmö

about 417

many-to-many case

POS tagging 307, 308, 309, 310, 311, 312

many-to-one case

Sentiment Analysis 300, 301, 303, 304, 305, 306, 307

marketplace 450

Markov property 411

Mask R-CNN Image Segmentatiog

reference link 146

mathematical tools, calculus

about 544

chain rule 548

derivatives 544

differentiation rules 548

gradient descent 546

gradients 544

matrix operations 548, 549

Matrix Multiply Unit (MMU) 574

matrix operations 548, 549

max-pooling operator 113

Mean Opinion Score (MOS) 179

metrics

Accuracy 19

Precision 19

Recall 19

reference link 19

Microsoft Azure

about 440, 441

reference link 440

Microsoft Azure Notebook

about 456

reference link 456

Microsoft Research Paraphrase Corpus (MRPC) 271

Mini-Batch Gradient Descent (MBGD) 564, 565

MIT License

reference link 514

ML Metadata (MLMD) 459

MNIST

reference link 15

MNIST dataset

PCA, implementing on 376, 377, 378

MNIST digits

generating, with DCGAN 200, 201, 204, 206, 207

used, for building GAN 193, 195, 196, 197, 198

MNIST digits classification example

reference link 478

MNIST (Modified National Institute of Standards and Technology)

logistic regression, using on 103, 104, 105, 107

TensorFlow Estimator API, using with 95, 96

mobile converter 463

mobile GPUs

reference link 473

mobile neural architecture search (MNAS) 470

mobile optimized interpreter 463

model

saving 68, 69

model free reinforcement learning 411

model generation 494

model optimization

reference link 462

multi-layered perceptron (MLP) 194

Multi-Layer Perceptron (MLP)

about 9

activation functions 14

ELU 13, 14

LeakyReLU 13, 14

ReLU 12

sigmoid function 11

solution, to problems 10, 11

tanh function 11

training, issues 10, 11

multiple linear regression 93

multiplicativ attention 330

multivariate linear regression 93

MuseNet

about 183

reference link 184

MxNet

reference link 1

N

natural language interface

reference link 525

natural language processing (NLP) 231

Natural Language Processing (NLP) 292

Neural Architecture Search (NAS) 494

neural embeddings

about 253

Item2Vec 254

node2vec 254, 256, 257, 259

neural networks

about 5, 7, 14

learnings 173, 174

Neural Style Transfer, with tf.keras

reference link 169

nightly build 587, 588

node2vec 254, 256, 257, 259

Not a Number (NaN) 285

N-series machines 451

N-series offerings

NC series 451

ND series 451

NV series 451

Nsynth

about 183

reference link 183

NSynth 179

Nsynth Colab

reference link 183

O

object detection

reference link 471

objective functions

binary_crossentropy 18

categorical_crossentropy 18

MSE 18

one-dimensional Convolutional Neural Network (1D CNN) 248

one-hot encoding

issue 232

limitations, overcoming 232

One hot-encoding (OHE) 15

one-to-many case

about 291

text tags, generating 292, 294, 295, 296, 297, 299

OpenAI gym

Breakout game, playing 419, 421, 422

reference link 418

OpenAI Gym

about 417, 418, 419

operations, TensorFlow 1.x

constants, declaring 55

examples 55

random tensors, creating 56, 57

sequences, generating 56

variables, creating 57, 58

optimizer learning rate

controlling 35

optimizers

reference link 18, 29

testing 28, 29, 30, 32, 34

output

predicting 47

P

paragraph embedding 263, 265

Paragraph Vectors - Distributed Bag of Words (PV-DBOW) 265

Paragraph Vectors - Distributed Memory (PV-DM) 265

paraphrase database (PPDB) 247

Part-of-Speech (POS) analysis 176

peephole LSTM 288, 289

perceptron

about 7, 8

placeholder

about 55

defining 58

plot command 383

policy-based methods 412

pooling layers, DCNN

about 113

average pooling 113

max pooling 113

pose estimation

reference link 471

Positive Rate (TPR) 507

POS tagging 307, 309, 310, 311, 312, 313, 317

prebuilt deep learning models

recycling, for extracting features 136, 137

pre-trained models, TensorFlow.js

BodyPix, reference link 486

Coco SSD, reference link 486

DeepLab v3, reference link 486

KNN Classifier, reference link 487

MobileNet, reference link 486

PoseNet, reference link 486

Speech Commands, reference link 487

Toxicity 487

Universal Sentence Encoder, reference link 487

pre-trained models, TensorFlow Lite

image classification 468, 470

mobile GPUs 473

object detection 468, 471

pose estimation 468, 471

question and answer 468, 472

reference link 468

segmentations 468, 471

smart reply 468, 471

style transfers 468, 471

text classification 468, 472

pretrained TPU models

using 584, 585

principal component analysis (PCA)

about 375, 376

implementing, on MNIST dataset 376, 377, 378

k-means clustering 380, 381

reference link 379

TensorFlow embeddings API 379

principal components 375

principles, reinforcement learning (RL)

goal 407

interaction, with environment 407

trial and error 407

Prioritized Experience Replay (PER) 415

PyTorch

reference link 1, 273

Q

quantization

about 462

post-training quantization 462

quantization-aware training 462

question and answer

reference link 472

R

ragged tensors 74

rainbow 436

Recurrent Neural Networks (RNN) 279

Region of Interest (ROI) 144

Region Proposal Network (RPN) 144

regression

about 87, 88

regularization

about 38

adoption, for avoiding overfitting 38, 39, 40

BatchNormalization 40

elastic regularization 40

L1 regularization (LASSO) 40

L2 regularization (Ridge) 40

regularizers

reference link 40

reinforcement learning (RL)

about 407, 409

achievements 415

action 409

model of the environment 411

policy 410

return 410

reward 410

state 409

value function 410

ReLU (REctified Linear Unit) 12

representational state transfer (REST)

reference link 509

residual networks 160

Restricted Boltzmann Machines (RBM) 375

about 393

backward pass 393

deep belief networks (DBNs) 397, 398

forward pass 392

used, for reconstructing images 393, 394, 396, 397

restriction 392

right singular matrix 376

RNN cell

about 280, 281, 282

backpropagation through time (BPTT) 283, 284

gradients, exploding 284, 285

gradients, vanishing 284, 285

variants 285

RNN cell variants

gated recurrent unit (GRU) 288

long short term memory (LSTM) 285, 286, 287

peephole LSTM 288, 289

RNN topologies

about 291, 292

many-to-many network 307, 308, 309, 310, 311, 313, 317

many-to-one network 300, 301, 302, 303, 304, 305, 306, 307

one-to-many network 292, 293, 294, 295, 296, 297, 298, 299

RNN variants

about 289

bidirectional RNNs 289, 290

stateful RNNs 290

Robot Operating System (ROS) 417

S

SageMaker

about 453

reference link 453

scaled dot-product attention 331

Scheduled Sampling 325

second-generation TPUs (TPU2) 576, 577

segmentation

reference link 471

Selective Search

reference link 143

self-organized maps (SOM)

about 384, 385, 386

used, for color mapping 387, 388, 389, 390, 391

sentence embeddings 263, 264

sentence vectors

generating 367, 369, 371, 372, 373, 374

sentiment analysis

about 44, 46, 47

seq2seq model (Encoder-Decoder architecture)

about 317, 318

machine translation example 319, 320, 322, 323, 324, 325, 327, 328

seq2seq network with attention

versus transformer network 337, 338

services, Microsoft Azure

Azure DevOps 442

Function 441

IoT Hub 441

Storage Services 441

Virtual Machines 441

shallow neural networks 392

Short Message Service (SMS) 243

simple linear regression 89, 90, 91, 92

simple Net, TensorFlow 2.0

improving, with Dropout 26, 28

improving, with hidden layers 23, 24, 25

singular value decomposition (SVD) 375

Skip-gram with Negative Sampling (SGNS) model 237

smart reply

reference link 471

soft update 415

Something2Vec page 254

SparkFun edge

reference link 464

sparse autoencoder

about 356, 357

reference link 357

SRGAN 209, 210

stacked autoencoder

about 362

Keras autoencoder example 367, 369, 371, 372, 373

stateful RNNs 290

state-of-the-art results, CIFAR-10

reference link 130

state-of-the-art results, MNIST

reference link 122

State-Value function 410

static embeddings

about 234

GloVe 238

Word2Vec 235

Stochastic Gradient Descent (SDG) 564, 565

Stochastic Gradient Descent (SGD) 21, 238

style transfer

reference link 471

style transfer

about 166, 167

content distance 167, 168

style distance 168

subword embeddings 260, 261

sum of squared error (SSE) 383

SVM-based classifier

reference link 143

T

Teacher Forcing 325

TensorFlow 459

about 569

MNIST handwritten digits, used for building GAN 193, 194, 196, 197, 198

reference link 108

TensorFlow 1.x

about 51

computational graph, program structure 51

constants 54

converting, to TensorFlow 2.x 80

operations, examples 55

placeholders 55

TensorFlow 1.x, example 59, 60

TensorFlow 2.0

changes 3, 4, 5

code example 8, 9

CycleGAN 218, 219, 220, 222, 223, 224, 225, 226, 227, 228

k-means clustering 381, 382, 383, 384

libraries, reference link 16

neural network, defining 16, 17, 18, 19, 21

optimizers, testing 28, 29, 30, 32, 34

simple Net, improving with Dropout 26, 28

simple Net, improving with hidden layers 23, 24, 25

TensorFlow 2.0 net

baseline, establishing 22

running 22

TensorFlow 2.1

using 587, 588

TensorFlow 2.x

about 60

Autograph 61, 62

callback 67

ConvNets 112

custom training 74, 75

distributed training 76

eager execution 60

ecosystem 81

Estimators 72, 73, 74

gradients, computing 74, 75

Keras APIs 63

libraries and extensions, reference link 82

model, saving 68, 69

namespaces, changes 79

native code, best practices 80, 81

ragged tensors 74

TensorFlow 1.x, converting to 80

tf.Keras 72, 73, 74

training, from tf.data.datasets 69, 70, 71, 72

using 80

weights, saving 68, 69

TensorFlow 2.x ecosystem

Accelerated Linear Algebra (XLA) 82

Colab 82

language bindings 82

MLPerf 82

Sonnet 81

TensorBoard 81

TensorBoard Federated 81

TensorBoard Playground 82

TensorBoard Probability 82

TensorFlow Datasets 82

TensorFlow Extended (TFX) 81

TensorFlow Hub 81

TensorFlow.js 81

TensorFlow Lite 81

TensorFlow Core r2.0

reference link 114

tensorflow-datasets

reference link 69

TensorFlow Data Validation (TFDV) 459

TensorFlow Enterprise

TensorFlow Estimator API

using, with MNIST 95, 96

TensorFlow Estimators

about 94

Canned Estimators 94

Custom Estimators 94

feature columns 94

feature columns, features 94

input functions 95

TensorFlow Extended (TFX)

about 439, 456, 457

libraries 459

pipeline 457, 458

pipeline components 458

reference link 81, 458

User Guide, reference link 459

TensorFlow federated (TTF) platform

federated core (FC) 476

federated learning (FL) 476

TensorFlow FL APIs

builders 477

datasets 477

models 476

TensorFlow Hub

about 159

Application Zoos, using with 158

eager execution, reference link 159

full integration , reference link 159

reference link 81, 159

TensorFlow.js

about 478

demo page, reference link 478

models, converting 485

Node.js 488

pre-trained models 485, 486, 487

reference link 81

Vanilla TensorFlow.js 478, 480, 481, 482, 483, 484

TensorFlow Keras layers

custom layers, defining 348, 349

TensorFlow Lite

about 462

accelerators, using 466

appplication example 466, 467, 468

appplication generic example 465, 466

architecture 464

FlatBuffers 463

GPUs, using 466

mobile converter 463

mobile optimized interpreter 463

pre-trained models, using 468

quantization 462

reference link 81

supported platforms 464

using 464

TensorFlow Lite converter

reference link 464

TensorFlow Metadata (TFMD) 459

TensorFlow Mobile

about 461

TensorFlow Model Analysis (TFMA) 459

TensorFlow Projector

reference link 478

TensorFlow Research Cloud (TFRC) 82

TensorFlow (TF)

about 1, 2

installing, reference link 4

reference link 1

TensorFlow Transform (TFT) 459

Tensor Processing Units (TPUs) 268

Tensor Processing Unit (TPU)

about 571, 572, 573

availability, checking 580

data, loading with tf.data 581

first-generation TPU (TPU v1) 574, 576

generations 573

model, building 581, 582, 584

model, loading into 581, 582, 584

performance 578, 579

second-generation TPUs (TPU2) 576, 577

third-generation TPUs (TPU3) 577, 578

using, with Colabs 580

Term Frequency-Inverse Document Frequency (TF-IDF) 232

text classification

reference link 472

Text-to-Speech (TTS) systems 179

textual documents

about 175, 176

tfjs-models

reference link 486

tf.Keras 72, 73, 74, 83, 84

Application Zoos, using with 158

used, for classifying Fashion-MNIST 147, 148, 149, 150

tf.Keras built-in VGG16 Net module

reference link 136

utilizing 135, 136

tf.lite API

reference link 465

TFLite Converter 464

TFLite FlatBuffer format 464

TFLite interpreter 464

third-generation TPUs (TPU3) 577, 578

TPUStrategy

reference link 78

transfer learning

Deep Inception-v3 Net, using 151, 153

used, for classifying horses and humans 154, 155, 157, 158

transformer architecture

about 337, 338, 339, 340

Transformers library

installation link 273

TTS systems

concatenative 180

parametric 180

U

U-Net

reference link 142

Unity ML-Agents SDK 417

Universal Language Model Fine Tuning (ULMFit) model

BERT, using as part of network 267

V

value-based methods 411

vanilla autoencoder

about 347

architecture 348

used, for reconstructing handwritten digits 350, 351, 353, 354

Vanilla TensorFlow.js 478, 479, 481, 482, 484

variables

about 54

creating 57

initializing 58

Variational Autoencoders (VAE) 399, 400, 401, 402, 403, 404

vectorization 232

Vector Processing Unit (VPU) 577

VGG16 Net

used, for recognizing cat image 134

videos

classifying, with pretrained nets 174, 175

virtual machine, on Microsoft Azure

creating 451

virtual machines

reference link 441

virtual machines (VM), on cloud

about 447

compute instance, on GCP 450

creating, on Amazon EC2 448

virtual machine, on Microsoft Azure 451

Visual Question Answering (VQA)

about 163, 165, 166

reference link 163

W

WaveNet 181, 182

about 179

reference link 179, 183

weights

saving 68, 69

Winner take all units (WTU) 384

Word2Vec

about 235, 236, 237, 247

architectures 235

CBOW architecture 235

reference link 237

skip-gram architecture 235, 236, 237

word embedding

about 231, 232

creating, gensim used 239, 240

space, exploring with gensim 240, 241, 242, 243

using, for spam detection 243, 244

word embedding, for spam detection

data, obtaining 244

data, processing 245, 246

matrix, building 247

model, evaluating 251

model, training 251

spam classifier, defining 248, 251

spam detector, running 252, 253

X

xception networks 161, 163

Y

YOLO network 145