Recurrent neural networks are a special breed of neural networks that are capable of reasoning over time. They are primarily used in scenarios where you have to deal with values that change over time.
In a regular neural network, you can provide only one input, which results in one prediction. This limits what you can do with a regular neural network. For example, regular neural networks are not good at translating text, while there have been quite a few successful experiments with recurrent neural networks in translation tasks.
In a recurrent neural network, it is possible to provide a sequence of samples that result in a single prediction. You can also use a recurrent neural network to predict an output sequence based on a single input sample. Finally, you can predict an output sequence based on an input sequence.
As with the other types of...
