TensorFlow 2.x in the Colaboratory Cloud: An Introduction to Deep Learning on Google’s Cloud Service

Table of Contents
About the Author
About the Technical Reviewer
Introduction
Chapter 1: Introduction to Deep Learning
	Neural Networks
	Learning Representation from Data
	TensorFlow 2.x
	Google Colab
	Google Drive
	Create a New Notebook
	GPU Hardware Accelerator
	Download a File from a URL
	Prepare the Dataset
	Colab Abends
	Colab Strange Results
	Tensors
	Scalars (0D Tensors)
	Vectors (1D Tensors)
	Matrices (2D Tensors)
	3D Matrices (3D Tensors)
	Key Attributes of Tensors
	Input Pipelines
	The tf.data API
	Function from_tensor_slices
	Iterate a tf.data.Dataset
	Tensors and numpy
	Chaining Transformations
	Mapping Tensors
	Filter a tf.data.Dataset
	Shuffling a Dataset
	TensorFlow Math
		Vector Tensors
		Matrix Tensors
		tf.data.Dataset Tensors
	Save a Notebook
	Download a Notebook to a Local Drive
	Load a Notebook from a Local Drive
Chapter 2: Build Your First Neural Network with Google Colab
	GPU Hardware Accelerator
	The load_digits Dataset
	Explore the Dataset
	Image Matrix
	Split Data into Train-Test Sets
	Build the Input Pipeline
	Explore TensorFlow Data
	Shuffle Data
	Continue Pipeline Construction
	Feedforward Neural Networks
	Number of Layers
	Our Model
	Model Summary
	Compile the Model
	Train the Model
	Model History
	Predictions
	Get an Image
	Mount Google Drive to Display an Image
Chapter 3: Working with TensorFlow Data
	TensorFlow Datasets (TFDS)
	Colab Abends
	Available TFDS
	Load a TFDS
	Extract Useful Information
	Inspect the TFDS
	Feature Dictionaries
	Build the Input Pipeline
	Build the Model
	Model Summary
	Compile and Train the Model
	Generalize on Test Data
	Visualize Performance
	DatasetBuilder (tfds.builder)
	MNIST Metadata
	Show Examples
	Prepare DatasetBuilder Data
	Build the Model
	Compile the Model
	Train the Model
	Generalize on Test Data
	Load CIFAR-10
	Inspect the Dataset
	Prepare the Input Pipeline
	Model the Data
Chapter 4: Working with Other Data
	Basic Mechanics
	TensorFlow Dataset Structure
	Reading Input Data
	Colab Abends
	Batch Size
	Keras Data
		Build the Input Pipeline
		Create the Model
		Model Summary
		Compile the Model
		Train the Model
	Scikit-Learn Data
		Explore the Data
		Build the Input Pipeline
		Build the Model
		Model Summary
		Compile the Model
		Train the Model
	Numpy Data
		Load Numpy Arrays with tf.data.Dataset
		Prepare Data for Training
		Create the Model
		Model Summary
		Compile the Model
		Train the Model
	Get Wine Data from GitHub
	CSV Data
		Dataset Characteristics
		Get Data
		Split Data into Train and Test Sets
		Prepare Data for TensorFlow Consumption
		Build the Model
		Model Summary
		Compile the Model
		Train the Model
	Get Abalone Data from GitHub
	Data Datasets
	Abalone Dataset
		Dataset Characteristics
		Mount Google Drive to Colab
		Read Data
		Explore Data
		Create Train and Test Sets
		Create Feature and Target Sets
		Scale Features
		Create Train and Test Sets with Sex and Scaled Values
		Convert Numpy Feature Sets into Pandas DataFrames
		Build the Input Pipeline
		Explore a Batch
		Categorical Columns
		Build the Model
		Compile the Model
		Train the Model
		Imbalanced and Irregular Data
		Dealing with Imbalanced Data
Chapter 5: Classification
	Fashion-MNIST Dataset
	Load Fashion-MNIST as a TFDS
		Explore the Dataset
		Build the Input Pipeline
		Build the Model
		Model Summary
		Compile the Model
		Train the Model
		Generalize on Test Data
		Visualize Performance
		Predict Labels for Test Images
		Build a Prediction Plot
	Load Fashion-MNIST as a Keras Dataset
		Explore the Data
		Visualize the First Image
		Visualize Sample Images
		Prepare Data for Training
		Build the Model
		Model Summary
		Compile the Model
		Train the Model
		Generalize on Test Data
		Visualize Training
		Predict Labels for Test Images
			Predict the First Image
			Predict Four Images
		Explore Misclassifications
		Visualize Misclassifications
		Predict from a Single Image
		Visualize Single Image Prediction
		Confusion Matrix
	Number of Hidden Layers
	Number of Neurons in Hidden Layers
Chapter 6: Regression
	Boston Housing Dataset
		Boston Data
		Explore the Dataset
		Create Feature and Target Sets
		Get Feature Names from the Features DataFrame
		Convert Features and Labels
		Split Dataset into Train and Test Sets
		Scale Data and Create TensorFlow Tensors
		Prepare Tensors for Training
		Create a Model
		Model Summary
		Compile the Model
		Train the Model
		Visualize Training
		Early Stopping
		Remove Bad Data
		Get Data
		Remove Noise
		Create Feature and Target Data
		Build the Input Pipeline
		Compile and Train
		Visualize
		Generalize on Test Data
		Make Predictions
		Visualize Predictions
		Load Boston Data from Scikit-Learn
		Remove Noise
		Build the Input Pipeline
		Model Data
	Model Cars Data
		Get Cars Data from GitHub
		Convert Categorical Column to Numeric
		Slice Extraneous Data
		Create Features and Labels
		Build the Input Pipeline
		Model Data
		Inspect the Model
		Visualize Training
		Generalize on Test Data
		Make Predictions
		Visualize Predictions
Chapter 7: Convolutional Neural Networks
	CNN Architecture
	Load Sample Images
	Display Images
	Scale Images
	Display Scaled Images
	Get More Images
	Mount Google Drive
	Copy Images to Google Drive
	Check Image Shapes
	Resize Images
	Create a Batch of Images
	Create Filters
	Plot Convolutional Kernels
	Apply a 2D Convolutional Layer
	Visualize Feature Maps
	CNN with Trainable Filters
	Building a CNN
		Load Data
		Display Information About the Dataset
		Extract Class Labels
		Display Samples
		Build a Custom Function to Display Samples
		Build a Custom Function to Display a Grid of Examples
		Pinpoint Metadata
		Build the Input Pipeline
		Create the Model
		Model Summary
		Model Layers
		Compile the Model
		Train the Model
		Generalize on Test Data
		Visualize Training Performance
		Predict Labels for Test Images
		Build a Prediction Plot
		Build a Custom Function
	Build a CNN with Keras Data
		Create Variables to Hold Train and Test Samples
		Display Sample Images
		Create the Input Pipeline
		Create the Model
		Compile and Train
		Predict
		Visualize Results
	Epilogue
Chapter 8: Automated Text Generation
	Natural Language Processing
	Generating Text with a RNN
		The Text File
		Mount Google Drive
		Read the Corpus into Memory
		Verify Corpus
		Create Vocabulary
		Vectorize the Text
		Create Integer Mappings
		Create Character Mappings
		Map a Sequence
		Vectorize the Corpus
		Predict the Next Character
		Create Training Input Sequences
		Display Samples
		Batch Sequences
		Create Samples and Targets
		Time Step Prediction
		Create Training Batches
		Build the Model
		Display Model Summary
		Check Output Shape
		Calculate Loss
		Compile the Model
		Configure Checkpoints
		Train the Model
		Rebuild the Model for Text Creation
			Restore Weights from Checkpoints
			Rebuild with Batch Size of 1
			Load Weights and Reshape
		Model Summary
		Create Components to Generate Text
			Create the Function
			Initialize Variables
			Vectorize and Reshape the Starting String
		Create New Text
Chapter 9: Sentiment Analysis
	IMDb Dataset
	Load IMDb as a TFDS
		Display the Keys
		Split into Train and Test Sets
		Display the First Sample
		Display Information About the TFDS
		Peruse Metadata
		Create the Encoder
		Use the Encoder on Samples
		Finish the Input Pipeline
		Create the Model
		Model Summary
		Compile the Model
		Train the Model
		Generalize on Test Data
		Visualize Training Performance
		Make Predictions from Fabricated Reviews
		Make Predictions on a Test Data Batch
		Prediction Accuracy for the First Batch
	Leverage Pretrained Embeddings
		Load the IMDb Dataset
		Build the Input Pipeline
		Create the Pretrained Model
		Compile the Model
		Train the Model
		Make Predictions
		Calculate Prediction Accuracy for the First Batch
	Explore IMDb with Keras
		Explore the Train Sample
		Create a Decoding Function
		Invoke the Decoding Function
		Continue Exploring the Training Sample
	Train Keras IMDb Data
		Load Data
		Pad Samples
		Build the Input Pipeline
		Build the Model
		Compile the Model
		Train the Model
		Predict
Chapter 10: Time Series Forecasting with RNNs
	Weather Forecasting
	The Weather Dataset
	Explore the Data
	Plot Relative Humidity Over Time
	Forecast a Univariate Time Series
		Scale Data
		Establish Training Split
		Create Features and Labels
		Create Train and Test Sets
		View Windows of Past History
		Plot a Single Example
		Create a Visual Performance Baseline
		Create a Baseline Metric
		Finish the Input Pipeline
		Explore a Data Window
		Create the Model
		Model Summary
		Verify Model Output
		Compile the Model
		Train the Model
		Generalize on Test Data
		Make Predictions
		Plot Model Performance
	Forecast a Multivariate Time Series
		Scale Data
		Multistep Model
		Generators
		Advantages of Using a Generator
		Generator Caveats
		Create a Generator Function
		Generate Train and Test Data
		Reconstitute Generated Tensors
		Finish the Input Pipeline
		Create the Model
		Model Summary
		Compile the Model
		Train the Model
		Generalize on Test Data
		Plot Performance
		Plot a Data Window
		Make a Prediction
Index