To discuss deep learning, we need to go back in time, not so long ago, to when big data arrived in front of all our faces. The term was, and still is, everywhere. It was a skill that everyone just had to have, a buzzword-compliant checklist item. But what exactly did that term really mean? Well, it just meant that rather than siloed SQL databases and files being FTP'ed to use, we had this explosion of digital data from social media, internet search engines, e-commerce sites, and much more. And of course, this data came in various forms and formats. More formally, we were suddenly dealing with unstructured data. Not only did we have this explosion of data due to applications such as Facebook, Twitter, Google, and more, but also the explosion continues. More and more people get and stay connected to each other, sharing vast amounts of information about themselves that they wouldn't dare provide to someone if asked via a telephone call, right? And we have little to no control over the format and quality of that data. This will become an important point as we proceed.

Now, this vast amount of data is great, but humans can barely absorb what they are exposed to daily, let alone this explosion of data. So, along the way, people realized that machine learning and artificial intelligence could be adapted to just such a task. From a simple machine learning algorithm, all the way up to multilayered networks, artificial intelligence and deep learning were born (at least the corporate world likes to believe it happened that way!).

Deep learning, which is a branch of machine learning and artificial intelligence, uses many levels of neural network layers (hierarchical, if you like) to perform its job. In many cases, these networks are built to mirror what we think we know about the human brain, with neurons connecting layers together like an intricately layered web. This allows data processing to occur in a nonlinear fashion. Each layer processes data from the previous layer (with an exception being the first layer, of course), passing its information on to the next layer. With any luck, each layer improves the model, and in the end, we achieve our goal and solve our problem.