Future Processing on deep learning's history

Deep Learning – how smart is it?

date: 5 May 2016
reading time: 2 min

As far as Artificial Intelligence (AI) is concerned, we associate it with know-it-all robots. However, it’s been fairly recently that AI really did become smart thanks to deep learning. How did it happen and how can we use this phenomenon to our advantage?

First things first, though. Let us tell you a little bit about a branch of AI, namely deep learning.

It is the software focused on simulating the activities going on in the layers of the brain – neurons in a neocortex, which distinguishes images, patterns, and data.

The evolution of Sci-fi brain

Sounds complex? Well, it is, and it wasn’t easy to achieve, as it took programmers almost 65 years to get to where they are now.

It all started with the development of neural networks in the 1950s when everything circled around simulations of the brain on a very basic level. Twenty years later, the network was able to differentiate patterns and simple images. The 1980s were the years when ‘deep’ models started to rise. The use of many layers of neurons enabled programmers to recognise speech and images. However, the process involved human touch and a lot of computing power.

True changes came in 2006 when the layers weren’t a problem anymore and could be effectively taught – each layer was dedicated to a different aspect of recognition from basic to more complicated ones.

What does it mean for us?

Much better speech and image recognition, that’s for sure. But which areas will improve the most? For now, the activities are still in the testing phase, however, the improvements are already visible.

Identification of objects has been tested in June 2015 by Google and appeared to be better than ever before. The system was able to distinguish specific objects and people in 10 million different YouTube videos.

Additionally, the advances led to better voice recognition on Android devices, which now makes far fewer mistakes.

It is also possible to transcribe any language with the error rate of just 7% which was demonstrated during Rick Rashid’s (Microsoft research officer) lecture in China.

As you can see, everything is going to be more…intelligent. There are many sectors that may benefit from these improvements like healthcare, transportation or communication. Researchers claim that AI is about to present us with significant changes and finally stop being associated only with Hogwarts reality.

Liked the blog post?

Please give us feedback.

Cookies

This website stores cookies on your computer. These cookies are used to improve our website and provide more personalized services to you, both on this website and through other media. To find out more about the cookies we use, see our Privacy policy.