Artificial Intelligence Programming with Python

CHAPTER 12Advanced AI Computing

“AI is neither good nor evil. It's a tool. It's a technology for us to use.”

—Oren Etzioni (American computer scientist)

CHAPTER OUTLINE

• 12.1 Introduction
• 12.2 AI with Graphics Processing Unit
• 12.3 AI with Tensor Processing Unit
• 12.4 AI with Intelligence Processing Unit
• 12.5 AI with Cloud Computing
• 12.6 Web-Based AI
• 12.7 Packaging the Code
• 12.8 AI with Edge Computing (Raspberry Pi, Google TPU Kit)
• 12.9 Create a Mobile AI App
• 12.10 Quantum AI
• 12.11 Summary
• 12.12 Chapter Review Questions

12.1 Introduction

Artificial intelligence algorithms are often computationally intensive and resource hungry; therefore, they are commonly run on graphic processing units (GPUs) and tensor processing units (TPUs). GPU and TPU hardware are often expensive; therefore, it is popular to utilize GPU and TPU capabilities through online platforms such as Google Colab and Kaggle.

Apart from running artificial intelligence algorithms on local computers, laptops, or desktops, artificial intelligence algorithms are also being increasingly run on remote servers; this is called cloud computing, or cloud AI. Whey they run on embedded devices, such as Raspberry Pi, microcontrollers, or your mobile phones, this is called edge computing, or edge AI.

Cloud AI and edge AI are getting more attention due to the fast development of the Internet of Things (IoT).

In this chapter, we will first introduce AI with GPUs and TPUs and then introduce cloud AI; we'll also show how ...