Deep Learning: An Introduction
What is Deep Learning?
In basic terms, Deep Learning is a type of Machine Learning which imitates the way that the human brain works in order to gather knowledge, teaching computers to perform tasks based on examples that come directly from images, text or sound. As opposed to traditional Machine Learning algorithms which follow a linear structure, Deep Learning algorithms are made up of neural networks inspired by the nerve cells that form the human brain. These networks contain many different layers, structured in a hierarchy which increases in complexity.
With Deep Learning, computers learn how to think for themselves, in a way that resembles how we as humans learn how to do so when we are young. For example, a child would learn what a cat is by pointing at things and saying cat, learning what is and what is not a cat by beginning to memorize and recognize a cat’s features. To teach a computer what a cat is, we would feed a set of training images of thousands of cats labelled ‘cat’ into the neural network. The computer would then identify cats based on patterns in behavior that it has recognized in the images.
Deep Learning in Our Day To Day Lives
Originally conceptualized in the mid-twentieth century, Deep Learning is not a new AI phenomenon, but in recent years, as machines have become more intelligent, it has really blown up, achieving results that are better than ever before, with deep neural networks even outperforming humans in some scenarios.
One of the companies at the forefront of DL technology is Facebook who have developed many applications and features based on Machine Learning, from their controversial facial recognition feature which notifies you when a picture of you that you have not been tagged in is uploaded, to their fake news alert button.
While some artificial intelligence inventions may seem unnecessary or even dystopic, there are many fields in which Deep Learning is having a massive impact on human lives, one of these being healthcare. An example of this is Project InnerEye, a Machine Learning tool that helps radiologists identify and analyze 3-D images of cancerous tumors. Another ground-breaking new DL technology that is set to impact the world is a collaboration between Google and Harvard University, which uses viscoelastic computations to predict earthquake aftershocks. This was achieved by training a neural network to look for patterns in a database of 131,000 mainshock-aftershock events, and then testing the predictions on a database of 30,000 similar pairs.
Deep Learning in the Future: How Deep is the Deep?
Although there is much debate surrounding DL and whether it is the future of AI or not, it is clear that the scope for future applications is massive, with Deep Learning technologies already advancing so much in a relatively short space of time. With companies like Microsoft, Google and Facebook investing millions into researching Deep Learning and neural networks, DL is going to become a bigger part of everyday lives, with it being predicted that Machine Learning will be used in most software applications in the near future.