Uses examples of sequential music and text generation to show how and when recurrent neural networks (specifically, long short-term memory networks) are useful.

Defines and gives some brief examples of neural networks.

Gives an example of using a neural network to predict one's salary based on a number of different characteristics and by using an activation function.

Explains how generative adversial neural networks create new data from existing data. The clip also contains an analogy of how a cashier and a counterfeiter are like the generative and discriminitory aspects of a generative adversarial network.

Explains how activation functions help make sense of data whilst feature generation groups data into its own categories.