Deep learning is a technique which teaches computers to do what we, as humans, do so naturally: learn by example. There are deep learning workstations available with tremendous capabilities too.

The car that doesn’t need a driver, the ability to know that a person is a person and not a bollard. The voice-controlled devices that we all enjoy so much. Hands-free speakers, tablets, TVs and your mobile phone. And still, deep learning is something that we all enjoy, yet know relatively little about.

And one of the critical reasons deep learning is so exciting is not just because they can take their instructions from the text, sound, and images. But, actually that they are beginning to perform better than humans. Their accuracy is higher, they are processing fasted, and that is exciting stuff.

How Does it Work?

This can get pretty complicated, depending on how technical you’d like to get. However, the very basics are:

-Most deep learning methods use Neural Networks
-You might hear deep learning referred to as deep neural networks
-Deep actually refers to the number of hidden layers in the neural network
-Deep networks can have as many as 150 layers, traditional; neural networks have 2-3 hidden layers
-Deep learning models are trained using vast sets of labeled data
-Machine Learning and Deep Learning are not the same things

What Is Machine Learning?

There is a basic definition of machine learning:

“Algorithms that parse data, learn from that data, and then apply what they’ve learned to make informed decisions.”

One of the easiest ways to understand this process is when it is applied to your music streaming service. You already have a playlist, and the program will use that playlist, combined with other things that you listened to, to make informed recommendations about what else you might like.

It uses algorithms to associate your preferences with other listeners who have similar preferences to create recommendations.

What is Deep Learning?

Deep learning is more analytic. It will take a range of data and begin to apply logic to it - and draw a conclusion. Which is much more human. Machines that are designed for deep learning need more information, which is where the neural network comes in.

There is a lot of training involved. If you are interested in playing with something that uses deep learning, then check out Google’s AlphaGo, and the 2017 Netlfix title of the same name.

Deep Learning Applications

It will come as no surprise that there are many applications for deep learning, now you understand a small amount about how it works. As mentioned above, here are some of the applications we will see:

-Self-driving cars
-Voice Search and Voice-Activated Assistants
-Automatic Machine Translations - given words in one language and translating them into another (see Google Translate App)
-Image Recognition
-Predicting Earthquakes
-Advertising

Take Away?

Over the next few years, more deep learning will be implemented by more companies and into more consumer technology. And no doubt we will see technology and innovations we can’t even imagine yet… probably made by the very same computers we are currently teaching.