The application of big data is enabling us to reveal deeper marketing insights all the time. Whether it’s open data revealing predictive patterns in consumer behaviour or hyper-personalisation creating a bespoke online shopping experience, big data in action is revolutionising the way we gather and act on business intelligence.
This in-depth data insight is made possible by machines; clever programs and algorithms that can process data at speeds no human could match, and derive meaning it would take us a lot longer to uncover. This is the power of artificial intelligence (AI), machine learning and deep learning; three industry buzzwords continuing to make a huge impact in the world of data marketing.
We’ve already taken a look at AI and how companies are using it to improve their services, and in this blog we’re delving into deep learning. What is deep learning and what could it help marketers achieve?
To define deep learning, we need to look at it in relation to AI and machine learning. Rather than separate entities, it might be helpful to think of AI, machine learning and deep learning as inter-connected; three stages of the same big data application. Although this is a simplification, for marketers it can help to signify the differences between these three related terms.
As we noted in our previous blog on artificial intelligence, AI can significantly shorten the distance between a need for information and the information itself; ‘thinking’ in a way humans do and performing the same tasks we can do, but often more quickly and efficiently.
Following on from this, we can think of machine learning as what happens behind the scenes of AI, the processes that enable machines to react to a question or request and locate the right information. By analysing huge data sets, machines essentially ‘learn’ to identify, differentiate and catalogue the various components of the information. Instead of telling a computer exactly how to do something, we tell it how to learn to do it.
One particular approach to machine learning is to use Artificial Neural Networks (ANN). For example, a shallow face detection program can be trained to detect the presence of a human face: The edges of the face are defined by the trainer from thousands of images classified as such. When given an image, the program recognises the angles and compositions of edges, and can then identify whether or not a human face is present.
So, where does deep learning come in? Deep learning is a subset of machine learning. Similar to shallow ANN, deep learning uses a series of binary questions to come to a decision. The biggest differences being that Deep Neural Networks (DNN) can be trained unsupervised and consist of many hidden layers, the advantage being that more abstract problems can be solved with less human intervention. For example, a DNN may try to learn low-level features, such as edges, in the early layers and then build upon this to learn more complex characteristics, such as nose shape and facial composition, in later layers. Where shallow ANN can learn to detect a human face, DNN can go further: detecting a face and then even recognising who the face belongs to, all in one end-to-end process.
This is when deep learning goes beyond human capabilities – true artificial intelligence. Deep learning is what many data marketers are really excited about, as rather than creating machines that can simply substitute what a human is capable of, a deep learning program can offer much more.
Much of what deep learning is thought to be able to do is yet to be realised. Relatively speaking, the technology is still in its infancy yet developing constantly, so all it may be capable of very much remains to be seen. Already though, the technology that could power uses such as self-driving cars or medicines custom-made to an individual’s genome – previously the stuff of science fiction – is on the horizon.
In terms of marketing, deep learning has the potential to help us find “patterns inside of patterns” (dmnews.com). As programs are exposed to more and more data, and become more adept at interpreting it and learning from it, marketers can turn to deep learning to reveal ever-more complex data relationships.
Deep-learning algorithms will bring order out of chaos, and hone in on data attributes like sentiment, emphasis and intent, the subtleties of human interaction that machines have so far failed to grasp. The marketing applications for this kind of intelligence are ever-growing: clustering and consumer classification; user preference recommendation systems; analysis of unstructured data (e.g. social media insight); advanced chatbots with personality; and reactive content generation are but a few.
Our blog on AI highlights some of the ways deep learning is already being put to use by some of the biggest names in tech, and these applications are continuing to develop at an astonishing rate. The only thing we can really be sure of is that deep learning is only just beginning to make its presence felt when it comes to marketing, so it’ll be fascinating to see where it takes us next.