How to Get Started With AI and Edge Computing

So far, I introduced you to the basic concepts of edge computing and AI, and I looked at a few alternatives to conventional edge computing hardware. In the last article, I discussed various free machine learning platforms. 

In this article, I’ll bring everything together and discuss what you need to get started if you have little to no experience with machine learning, AI, and edge computing.

Edge Computing Hardware

As discussed in A Look at Edge Computing and AI HATs and Shields, edge computing brings some calculations closer to the user and away from the cloud itself. So the hardware that performs the calculations sits somewhere between the user’s network and the cloud.

Therefore, you can essentially use any computer that’s able to connect to your existing network as well as to the cloud.

Many tasks can also be performed by single-board computers, like the Raspberry Pi. More demanding tasks, however, might require a dedicated server that can handle the additional load. 

Therefore, you can essentially use any computer that’s able to connect to your existing network as well as to the cloud. 

Many tasks can also be performed by single-board computers, like the Raspberry Pi. More demanding tasks, however, might require a dedicated server that can handle the additional load. 

If you’re a hobbyist or maker, a Raspberry Pi with a dedicated co-processor HAT should be more than sufficient to perform everyday tasks. However, various solutions for businesses and enterprise users exist. Those mainly intend to be easy to maintain and cheap to run.

AI Basics for Beginners

If you are completely new to the field of AI and machine learning, I recommend you start by studying the basics of artificial intelligence. 

There are many great resources out there, and a lot of them are free. However, I can recommend that you take a look at the book Artificial Intelligence: A Modern Approach to get a good overview of the topic. You can also enroll in an online course if you’re not that into reading. However, be aware that they might be quite expensive.

Hardware for Machine Learning and AI

Once you know the basics, it’s time to think about the hardware. The good thing is that most modern hardware can be used for machine learning and AI.

However, to boost the performance, you can take a look at dedicated hardware that was specifically made for this task. The Jetson Nano is a popular SBC that can be used for all sorts of AI applications.

Also, if you plan on using TensorFlow, you should take a look at a TPU that supports your hardware. Those can be cloud-based, but offline alternatives, like the Coral USB Accelerator, as well as dedicated single-board computers, are available.

Software for Machine Learning and AI

Unfortunately, there’s no single go-to software solution for all AI and machine learning applications. 

However, you can take a look at this summary of popular free ML and AI platforms. Note that you’ll be limited by the hardware you choose. For example, if you plan on using a Raspberry Pi, you’ll only be able to use software that can run on it, and you can only use programming languages that are supported.

Data Collection for Machine Learning

Some learning methods require you to use a large set of examples and expected results. You can either collect the data yourself, or you can use a database that contains many different examples and large data-sets. OpenML is a free website that allows users to upload machine learning data and results from experiments.

So, What Do You Need for AI and Edge Computing?

Hobbyists and makers should use the hardware that they already own because most AI platforms can be utilized on almost every modern computer. 

An external co-processor can drastically speed up the learning process and calculations. Businesses and enterprise users should evaluate what tasks they want to perform since they might need more powerful hardware. 

The software that can be used is mainly limited by the chosen platform and the task itself.