Designed by humans, traditional intelligent algorithms depends greatly on experience and sometimes even luck. Nowadays, it is possible to get machines to automatically learn some of the features- the main objective of Artificial Intelligence (AI).

Deep learning is intrinsically different from other algorithms as it solves insufficiencies of traditional algorithms in the following aspects.

Unlike the two 3 layered structures of traditional algorithms, deep learning has a much deeper structure. With the number of layers reaching over a hundred, large amounts of data in complex classifications can be processed.

In essence, the higher the level, the more specific the components. Just like human brain, an original signal passes through layers upon layers of processing. It then takes a partial understanding (shallow) to an overall abstraction (deep) where we perceive the object.