Within the family of computerized learning, you might think of “deep learning” as the gifted child of machine learning. Climb further up the family tree, and you’ll find Artificial Intelligence as the revered grandfather of them all.
Here’s how it breaks down.
At the top of the hierarchy is Artificial Intelligence (AI). AI is designed to perform tasks as programmed, i.e., it is fed a sophisticated algorithm and programming and is let loose on things like massive data sets. AI can be trained to find patterns, solve puzzles, improve efficiencies, and so on.
Below AI you will find “machine learning,” which derives from AI. As the AI gathers and analyzes huge amounts of information, it can use new information to learn and refine its knowledge of a process and its execution of a task. Over time, the algorithm gets ‘smarter.’ In essence, machine learning focuses on solving real-world problems with neural networks designed to mimic a super-human’s own decision-making.
Deep learning is the final tier in this hierarchy, but also the most ambitious. Deep learning derives from machine learning but can be compared to you learning something new on your own. Through its own algorithm and computing work, deep learning is essentially using its own brain, known as its “Deep Neural Network,” to solve just about any problem which requires “thought” – human or artificial.
These deep neural networks are designed to operate just as the neural networks found in the human brain. These networks – logical constructions which ask a series of binary true/false questions, or extract a numerical value, of every bit of data which pass through them—can classify information according to the answers received. In this sense, deep learning involves feeding a computer system a lot of data, which it can use to make decisions about other data.
Deep learning can be applied to any form of data – machine signals, computer vision, audio, video, social network filtering, bioinformatics and drug design, speech, written words, the list goes on – to produce nearly immediate conclusions that, to the unknowing eye, probably seem as if they had been arrived at by humans who spent a lot of time thinking about the conclusion.
Where is Deep Learning being Used?
Deep learning is used by Google in its voice and image recognition algorithms, by Netflix and Amazon to decipher your interests and figure out what you should watch or buy next, and even by researchers at MIT to try and predict the future.
Where deep learning is likely to have the most immediate impact is in the field of autonomous, self-driving vehicles. As the vehicles make more trips and store more experiences (data), it will apply that new data to becoming an even better driver, and avoiding the same mistakes twice (unlike humans).
But there are so many more and endless uses for deep learning. A system recently developed by a team of British and American researchers was shown to be able to correctly predict a court’s decision, once it was fed the basic facts of the case. Deep learning will also be the driver of gains in precision medicine, as it can better understand which medicines work better for a very specific version of a disease on a very particular type of person. The list goes on.