A New Way To Learn

Among the rising field of machine learning, new techniques such as deep learning have captured the attention of programmers and researchers alike. This begs the question: what is deep learning and why is it so important?

To recap, machine learning involves computers analyzing existing data to find trends and attempt to discover the cause-and-effect relationship between multiple variables. This involves a long process called “training”. This is when a computer is left to run a ML program for an extended period of time in hopes of finding these patterns. The research done in this field focuses most extensively on manufacturing the algorithms involved with finding patterns most efficiently, depending on the data provided.

Machine learning methods rely on neural networks, which are series of nodes (or units) that connect to one another. This system is inspired by the neural networks found in animal brains that produce a sensation of thoughts and realization. In short, this system allows us to think!

Deep learning is a method that uses representation learning. Otherwise known as feature learning, this technique involves automatically detecting the variables or “features” to find in data. For some data, this isn’t necessary. Early experiments with self-driving cars, for example, were straightforward in one respect: we knew the variables. While driving, the average person looks for certain cues and reacts to them. We can train an AI to look for road markings and stop signs, and we know how we are supposed to react to them.

Deep learning is so powerful because it allows us to look at tasks that humans can’t achieve and show us the cause-and-effect relationships that aren’t immediately apparent. Not only could this perform the calculations that we need, it could suggest associations we never noticed in the first place, furthering our understanding of science in every field.

Deep Learning using Neural Networks

Picture Source: semiengineering.com