Deep Learning, Machine Learning, and Artificial Intelligence – What are the Differences?

In this video, Bernardo F. Nunes explains how these 3 concepts (artificial intelligence, machine learning, and deep learning) do not represent the same thing:

Bernardo is breaking down the 3 concepts for us, in a very easy and understandable way.

Artificial intelligence exists when a machine has cognitive capabilities, such as problem solving and learning. It’s normally associated to a human benchmark, as, for example reasoning, speech, and vision.

In the video, he differentiates 3 levels of A.I.:

1. Narrow A.I., when a machine is better than us in a specific task (we are here now)
2. General A.I., when a machine is like us in any intellectual task
3. Strong A.I, when a machine is better than us in many tasks

One of the favorites early developments in A.I. is the perceptron (1957). It was a single layer of artificial neural networks designed for image recognition. They are called neural networks because the first practitioners on A.I. thought that these interconnected nodes looked like the human neural system, which has neural networks. These are the natural neural networks. The perceptron is a rudimentary version of an artificial neural network.

Machine learning, appeared on the 1980s, when a body of researchers worked on what is called supervised learning. Algorithms are trained with datasets based on past examples, in a model in which the trained algorithm is applied to a new dataset for classification purposes. Widely used for business purposes.

Deep learning makes use of deep neural networks. Shallow neural nets have only one hidden layer between the input and the output. However, deep neural nets have 2 or more hidden layers between the input and the output. It’s the responsible for the advancement in image recognition. If you can represent an image numerically, then you can process it with deep learning.

Bernardo’s conclusion is that deep learning, machine learning, and artificial intelligence are not 3 different things. They simply are subsamples of each other: deep learning belongs to machine learning, and machine learning belongs to artificial intelligence.