Book Image

Hands-On Python Deep Learning for the Web

By : Anubhav Singh, Sayak Paul
Book Image

Hands-On Python Deep Learning for the Web

By: Anubhav Singh, Sayak Paul

Overview of this book

When used effectively, deep learning techniques can help you develop intelligent web apps. In this book, you'll cover the latest tools and technological practices that are being used to implement deep learning in web development using Python. Starting with the fundamentals of machine learning, you'll focus on DL and the basics of neural networks, including common variants such as convolutional neural networks (CNNs). You'll learn how to integrate them into websites with the frontends of different standard web tech stacks. The book then helps you gain practical experience of developing a deep learning-enabled web app using Python libraries such as Django and Flask by creating RESTful APIs for custom models. Later, you'll explore how to set up a cloud environment for deep learning-based web deployments on Google Cloud and Amazon Web Services (AWS). Next, you'll learn how to use Microsoft's intelligent Emotion API, which can detect a person's emotions through a picture of their face. You'll also get to grips with deploying real-world websites, in addition to learning how to secure websites using reCAPTCHA and Cloudflare. Finally, you'll use NLP to integrate a voice UX through Dialogflow on your web pages. By the end of this book, you'll have learned how to deploy intelligent web apps and websites with the help of effective tools and practices.
Table of Contents (19 chapters)
Artificial Intelligence on the Web
Using Deep Learning for Web Development
Getting Started with Different Deep Learning APIs for Web Development
Deep Learning in Production (Intelligent Web Apps)
Appendix: Success Stories and Emerging Areas in Deep Learning on the Web

DIY - A Web DL Production Environment

In previous chapters, we saw how to use some notable Deep Learning (DL) platforms, such as Amazon Web Services (AWS), Google Cloud Platform (GCP), and Microsoft Azure, to enable DL in our web applications. We then saw how to make websites secure using DL. However, in production, the challenge is often not just building the predictive model—the real problems arise when you want to update a model that is already sending responses to users. How much time and business can you lose in the 30 seconds or 1 minute that it may take to replace the model file? What if there are models customized for each user? That might even mean billions of models for a platform such as Facebook.

You need to have definite solutions for updating models in production. Also, since the ingested data may not be in the format that the training is performed in, you...