Deep learning is a term that many may be oblivious to, but it’s one set to have a tremendous impact on our lives in upcoming years.
Deep Learning algorithms are one of the most valuable tools for making sense of Big Data. Deep learning is basically what happened when machine learning and Big Data intersected. It is an approach to building and training neural networks that involves using a set of generative, hierarchical learning mechanisms to autonomously generate high-level representations from raw and unsupervised data sources. It then uses these representations to carry out typical machine learning tasks, such as classification and clustering.
One of its most widely used applications is image recognition, which perhaps provides the simplest explanation for how it works. Using data learning algorithms, a large number of images are collected - faces are a common example. It then learns all of these images as faces. It also collects a large quantity of pictures that are not faces for comparison, learning these as things that aren’t faces, although there is some debate around whether this part of the process is necessary. Once the machine has learned what an image portrays using this method, it is able to make an accurate prediction as to what an image is based on the data analysed (in this case pixels) and label it. There are a number of other everyday applications for deep learning. Speech recognition is one that many will use on an everyday basis, especially Android Phone users, with the speech recognition software heavily relying on data learning.
Deep learning has really snowballed of late because of its implications for AI, and the substantial investment that the field is subsequently seeing from the large tech firms, particularly Google. Google Brain is the web giant’s deep learning research project, and is one of the main ways in which Google is innovating in the field. They have also been heavily engaged in a talent grab, purchasing companies such as Britain’s DeepMind Technologies in 2014 for a figure believed to be in the region of £242m. Deep Learning is likely to be behind every technology that humans perceive to be ‘magical’ and mystifying in the next few years. Its association with AI is, however, one of the reasons it has come under extra scrutiny, with criticism coming from those paranoid about excessive computer intelligence and the singularity.
DeepMind is a good example of where deep learning is heading, and why it will be so important in the years to come. DeepMind claims that their system is not pre-programmed and learns from experience, meaning that it can be self-sustaining. Their AI can, for example, learn how to play a video game with no change to the code, and often plays it at a higher level than human beings are capable of.
Deep learning has a number of applications that are useful for business. Other than the clear benefits from AI, deep learning makes available to firms all of the information available in the massive datasets of Big Data to exploit. It has also proven useful in other areas of business. A number of companies have seen success using deep reinforcement learning in direct marketing settings with its application for CRM automation. A neural network was used to roughly establish the value of possible direct marketing actions over the customer state space, defined in terms of Recency, Frequency and Monetary (RFM) variables, which was revealed to have a natural interpretation as CLV (customer lifetime value).
Danny Hillis wrote in his book 'The Pattern On The Stone' that ‘the greatest achievement of our technology may well be the creation of tools that allow us to go beyond engineering - that allow us to create more than we can understand.’ Its implications for business are many, and transferring it into action is vital.