Hands-On Python Deep Learning for the Web: Integrating neural network architectures to build smart web apps with Flask, Django, and TensorFlow

Cover
Title Page
Copyright and Credits
About Packt
Contributors
Dedication
Preface
Table of Contents
Section 1: Artificial Intelligence on the Web
Chapter 1: Demystifying Artificial Intelligence and Fundamentals of Machine Learning
	Introduction to artificial intelligence and its types
		Factors responsible for AI propulsion
			Data
			Advancements in algorithms
			Advancements in hardware
			The democratization of high-performance computing
	ML – the most popular form of AI
	What is DL?
	The relation between AI, ML, and DL
	Revisiting the fundamentals of ML
		Types of ML
			Supervised learning
			Unsupervised learning
			Reinforcement learning
			Semi-supervised learning
		Necessary terminologies
			Train, test, and validation sets
			Bias and variance
			Overfitting and underfitting
			Training error and generalization error
	A standard ML workflow
		Data retrieval
		Data preparation
			Exploratory Data Analysis (EDA) 
			Data processing and wrangling
			Feature engineering and extraction/selection
		Modeling
			Model training
			Model evaluation
			Model tuning 
		Model comparison and selection
		Deployment and monitoring
	The web before and after AI
		Chatbots
		Web analytics
		Spam filtering
		Search
	Biggest web-AI players and what are they doing with AI
		Google
			Google Search
			Google Translate
			Google Assistant
			Other products
		Facebook
			Fake profiles
			Fake news and disturbing content
			Other uses
		Amazon
			Alexa
			Amazon robotics
			DeepLens
	Summary
Section 2: Using Deep Learning for Web Development
Chapter 2: Getting Started with Deep Learning Using Python
	Demystifying neural networks
		Artificial neurons
			Anatomy of a linear neuron
			Anatomy of a nonlinear neuron
		A note on the input and output layers of a neural network
		Gradient descent and backpropagation
	Different types of neural network
		Convolutional neural networks
		Recurrent neural networks
			Feeding the letters to the network
			Initializing the weight matrix and more
			Putting the weight matrices together
			Applying activation functions and the final output
	Exploring Jupyter Notebooks
		Installing Jupyter Notebook
			Installation using pip
			Installation using Anaconda
		Verifying the installation
		Jupyter Notebooks
	Setting up a deep-learning-based cloud environment
		Setting up an AWS EC2 GPU deep learning environment
			Step 1: Creating an EC2 GPU-enabled instance
			Step 2: SSHing into your EC2 instance
			Step 3: Installing CUDA drivers on the GPU instance
			Step 4: Installing the Anaconda distribution of Python
			Step 5: Run Jupyter
		Deep learning on Crestle
		Other deep learning environments
	Exploring NumPy and pandas
		NumPy
			NumPy arrays
			Basic NumPy array operations
			NumPy arrays versus Python lists
				Array slicing over multiple rows and columns
				Assignment over slicing
		Pandas
	Summary
Chapter 3: Creating Your First Deep Learning Web Application
	Technical requirements
	Structuring a deep learning web application
		A structure diagram of a general deep learning web application
	Understanding datasets
		The MNIST dataset of handwritten digits
		Exploring the dataset
			Creating functions to read the image files
			Creating functions to read label files
			A summary of the dataset
	Implementing a simple neural network using Python
		Importing the necessary modules
		Reusing our functions to load the image and label files
		Reshaping the arrays for processing with Keras
		Creating a neural network using Keras
		Compiling and training a Keras neural network
		Evaluating and storing the model
	Creating a Flask API to work with server-side Python
		Setting up the environment
		Uploading the model structure and weights
		Creating our first Flask server
		Importing the necessary modules
		Loading data into the script runtime and setting the model
		Setting the app and index function
		Converting the image function
		Prediction APIs
	Using the API via cURL and creating a web client using Flask
		Using the API via cURL
		Creating a simple web client for the API
	Improving the deep learning backend
	Summary
Chapter 4: Getting Started with TensorFlow.js
	Technical requirements
	The fundamentals of TF.js
		What is TensorFlow?
		What is TF.js?
		Why TF.js?
		The basic concepts of TF.js
			Tensors
			Variables
			Operators
			Models and layers
		A case study using TF.js
			A problem statement for our TF.js mini-project
			The Iris flower dataset
	Your first deep learning web application with TF.js
		Preparing the dataset
		Project architecture
		Starting up the project
		Creating a TF.js model
		Training the TF.js model
		Predicting using the TF.js model
		Creating a simple client
		Running the TF.js web app
	Advantages and limitations of TF.js
	Summary
Section 3: Getting Started with Different Deep Learning APIs for Web Development
Chapter 5: Deep Learning through APIs
	What is an API?
	The importance of using APIs
	How is an API different from a library?
	Some widely known deep learning APIs
	Some lesser-known deep learning APIs
	Choosing a deep learning API provider
	Summary
Chapter 6: Deep Learning on Google Cloud Platform Using Python
	Technical requirements
	Setting up your GCP account
	Creating your first project on GCP
	Using the Dialogflow API in Python
		Creating a Dialogflow account
		Creating a new agent
		Creating a new intent
		Testing your agent
		Installing the Dialogflow Python SDK
		Creating a GCP service account
		Calling the Dialogflow agent using Python API
	Using the Cloud Vision API in Python
		The importance of using pre-trained models
		Setting up the Vision Client libraries
		The Cloud Vision API calling using Python
	Using the Cloud Translation API in Python
		Setting up the Cloud Translate API for Python
		Using the Google Cloud Translation Python library
	Summary
Chapter 7: DL on AWS Using Python: Object Detection and Home Automation
	Technical requirements
	Getting started in AWS
	A short tour of the AWS offerings
	Getting started with boto3
		Configuring environment variables and installing boto3
		Loading up the environment variables in Python
		Creating an S3 bucket
		Accessing S3 from Python code with boto3
	Using the Rekognition API in Python
	Using the Alexa API in Python
		Prerequisites and a block diagram of the project
		Creating a configuration for the skill
		Setting up Login with Amazon
		Creating the skill
		Configuring the AWS Lambda function
		Creating the Lambda function
		Configuring the Alexa skill
		Setting up Amazon DynamoDB for the skill
		Deploying the code for the AWS Lambda function
		Testing the Lambda function
		Testing the AWS Home Automation skill
	Summary
Chapter 8: Deep Learning on Microsoft Azure Using Python
	Technical requirements
	Setting up your account in Azure
	A walk-through of the deep learning services provided by Azure
	Object detection using the Face API and Python
		The initial setup
		Consuming the Face API from Python code
	Extracting text information using the Text Analytics API and Python
		Using the Text Analytics API from Python code
	An introduction to CNTK
		Getting started with CNTK
			Installation on a local machine
			Installation on Google Colaboratory
		Creating a CNTK neural network model
		Training the CNTK model
		Testing and saving the CNTK model
	A brief introduction to Django web development
		Getting started with Django
		Creating a new Django project
		Setting up the home page template
	Making predictions using CNTK from the Django project
		Setting up the predict route and view
		Making the necessary module imports
		Loading and predicting using the CNTK model
		Testing the web app
	Summary
Section 4: Deep Learning in Production (Intelligent Web Apps)
Chapter 9: A General Production Framework for Deep Learning-Enabled Websites
	Technical requirements
	Defining the problem statement
		Building a mental model of the project
		Avoiding the chances of getting erroneous data in the first place
	How not to build an AI backend
		Expecting the AI part of the website to be real time
		Assuming the incoming data from a website is ideal
	A sample end-to-end AI-integrated web application
		Data collection and cleanup
		Building the AI model
			Making the necessary imports
			Reading the dataset and preparing cleaning functions
			Slicing out the required data
			Applying text cleaning
			Splitting the dataset into train and test parts
			Aggregating text about products and users
			Creating TF-IDF vectorizers of users and products
			Creating an index of users and products by the ratings provided
			Creating the matrix factorization function
			Saving the model as pickle
	Building an interface
		Creating an API to answer search queries
		Creating an interface to use the API
	Summary
Chapter 10: Securing Web Apps with Deep Learning
	Technical requirements
	The story of reCAPTCHA
	Malicious user detection
	An LSTM-based model for authenticating users
		Building a model for an authentication validity check
		Hosting the custom authentication validation model
	A Django-based app for using an API
		The Django project setup
		Creating an app in the project
		Linking the app to the project
		Adding routes to the website
		Creating the route handling file in the billboard app
		Adding authentication routes and configurations
		Creating the login page
		Creating a logout view
		Creating a login page template
		The billboard page template
		Adding to Billboard page template
		The billboard model 
		Creating the billboard view 
		Creating bills and adding views
		Creating the admin user and testing it
	Using reCAPTCHA in web applications with Python
	Website security with Cloudflare
	Summary 
Chapter 11: DIY - A Web DL Production Environment
	Technical requirements
	An overview of DL in production methods
		A web API service
		Online learning
		Batch forecasting
		Auto ML
	Popular tools for deploying ML in production
		creme
		Airflow
		AutoML
	Implementing a demonstration DL web environment
		Building a predictive model
			Step 1 – Importing the necessary modules
			Step 2 – Loading the dataset and observing
			Step 3 – Separating the target variable
			Step 4 – Performing scaling on the features
			Step 5 – Splitting the dataset into test and train datasets
			Step 6 – Creating a neural network object in sklearn
			Step 7 – Performing the training
		Implementing the frontend
		Implementing the backend
	Deploying the project to Heroku
	Security measures, monitoring techniques, and performance optimization
	Summary
Chapter 12: Creating an E2E Web App Using DL APIs and Customer Support Chatbot
	Technical requirements
	An introduction to NLP
		Corpus
		Parts of speech
		Tokenization
		Stemming and lemmatization
		Bag of words
		Similarity
	An introduction to chatbots
	Creating a Dialogflow bot with the personality of a customer support representative
		Getting started with Dialogflow
			Step 1 – Opening the Dialogflow console
			Step 2 – Creating a new agent
			Step 3 – Understanding the dashboard
			Step 4 – Creating the intents
				Step 4.1 – Creating HelpIntent
				Step 4.2 – Creating the CheckOrderStatus intent
			Step 5 – Creating a webhook
			Step 6 – Creating a Firebase cloud function
				Step 6.1 – Adding the required packages to package.json
				Step 6.2 – Adding logic to index.js
			Step 7 – Adding a personality to the bot
	Using ngrok to facilitate HTTPS APIs on localhost
	Creating a testing UI using Django to manage orders
		Step 1 – Creating a Django project
		Step 2 – Creating an app that uses the API of the order management system
		Step 3 – Setting up settings.py
			Step 3.1 – Adding the apiui app to the list of installed apps
			Step 3.2 – Removing the database setting
		Step 4 – Adding routes to apiui
		Step 5 – Adding routes within the apiui app
		Step 6 – Creating the views required
			Step 6.1 – Creating indexView
			Step 6.2 – Creating viewOrder
		Step 7 – Creating the templates
	Speech recognition and speech synthesis on a web page using the Web Speech API
		Step 1 – Creating the button element
		Step 2 – Initializing the Web Speech API and performing configuration
		Step 3 – Making a call to the Dialogflow agent
		Step 4 – Creating a Dialogflow API proxy on Dialogflow Gateway by Ushakov
			Step 4.1 – Creating an account on Dialogflow Gateway
			Step 4.2 – Creating a service account for your Dialogflow agent project
			Step 4.3 – Uploading the service key file to Dialogflow Gateway
		Step 5 – Adding a click handler for the button
	Summary
Appendix: Success Stories and Emerging Areas in Deep Learning on the Web
Other Books You May Enjoy
Index