Deep learning is part of the broader machine learning methods. Supervised unsupervised or semi-supervised.

Recently, over the last couple of years the methods in this class have become increasing popular as they allow for design of systems that go beyond that possible with traditional machine learning methods.

Good for complex analyses where the system is required to learn features and there is a lot of data available for training, or existing training results can be applied

More traditional methods, linear analysis, SVM, boosting, random forests.

Can be easier to use traditional machine learning methods where the amount of data is more limited, has limited dimensionality or extremely noisy.