It's Alive!!!

30th July 2015

Digital advancements are coming thick and fast. You get your head round the latest technology only to realise that once again you are the kid with the old football boots while everyone else has picked up the latest ones.

Well have no fear! In this post, we’re going to give you the skinny on deep learning so you can be the kid in the new football boots (or equivalent sports attire!).

Deep learning is not necessarily a new thing but in 2015 it is being seen as one of the hot topics and is gaining traction quickly. Companies like Microsoft, Google and Facebook are investing heavily in the technology and its inherent power provides some exciting opportunities for a range of applications.

For years, scientists have engaged in a battle of one-upmanship with each other in the race to develop artificial intelligence. There’s a whole raft of theories but at the heart of it is the theory that human intelligence stems from a single algorithm. This in turn has pushed forward a movement that seeks to meld computer science with neuroscience, developing neural networks that mimic the behaviour of the human brain.

What is deep learning?

Deep learning is a powerful tool. For a start, it trains machines to recognise patterns in the data so they can be classified and categorised without human interaction. And, perhaps more importantly, it gives us the ability to access and manipulate huge reams of data that were previously inaccessible.

For the nervous few (we’re looking at you Sarah Connor), this is a sign to shore up in a nuclear bunker awaiting the machine-driven apocalypse. For the sane majority, deep learning opens a world of possibility.

Why should I care?

A lot happens on the internet and there’s a whole world of toys for marketers and digital teams. Deep learning has an impact on each of these and it’s why the big boys are investing heavily in the technology so they can deliver even better user experiences.

We have already seen some applications of deep learning in the tools we use every day, e.g. Netflix, Amazon, mobile assistants, Google Search, Gmail, Facebook and Twitter. Powerful recommendation tools and advertising tools are harnessing the power of deep learning to deliver the best possible content.

But what’s next...

  • Automated translation is often a bone of contention. Tools like Google Translate can give you the gist of what is being said but lack the context to be able to translate content effectively. Studies in automatic speech recognition and natural language processing are working towards closing that gap, enabling machines to understand languages as we would and producing closer translations.
  • Large-scale image recognition is also developing quickly. Automatic image captioning and tagging is highly dependent on the machine understanding the image. At the moment, we rely heavily on alt-text and other describers to try and provide context to the machine. However, websites are becoming more graphical in a bid to attract Millenials who, according to recent studies, are drawn to sites with compelling visual elements and graphics. Image recognition driven by deep learning will help us to push the graphical boundaries while retaining that all important search ranking.
  • Google Maps are using their DeepStereo algorithm to synthesize real world images, giving us clearer indications on travelling and exploring. The algorithm harnesses deep learning to compare certain frames used by Google Maps and insert the missing frames required to provide that immersive experience.
  • Search Engine Optimisation is an important consideration for any business and deep learning is being used to enhance SEO and analytics, with particular attention to the reduction of spam sites and increased device integration. Even the controversial “supercookies” can be used in conjunction with deep learning to develop even greater profiles on your users.
  • Deep learning is also being introduced into CRM systems to support and enhance marketing activities alongside a greater understanding of your customers from all manner of angles.

I've started so I'll finish...

The use of deep learning within the digital arena is increasing and the initial attempts have been promising. There’s been a few hiccups along the way (ahem, Google) but over the next few years you can expect to see a greater adoption of the technology in a lot of the tools and websites you use.

Rich Madigan is a Project Manager and Kentico Consultant here at MMT Digital - a mind of (hopefully) useful information.

The original article can be found at