Just like CNNs were specially designed to process images, Recurrent Neural Networks (RNNs) are specially designed to process sequential data.
In sequential data, entities occur in a particular order. If you break the order, you don’t have a meaningful sequence anymore. For example, you could have a sequence of words which makes up a document. If you jumble the works, you will end up having a nonsensical document. Similarly, you could have a sequence of images which makes up a video. If you shuffle the fames, you’ll end up having a different video. Likewise, you could have a piece of music which comprises of a sequence of notes. If you change the notes, you’ll mess up the melody.
Recurrent neural networks are variants of the vanilla neural networks which are tailored to learn sequence patterns, Lets understand sequences from professor Raghavan.
You saw some examples of sequence problems. Let’s now hear an interesting unconventional example of a problem involving sequences which can be solved using RNNs.
Although sorting is a problem that involves an algorithm, but the fact that RNNs can learn an algorithm speaks volume about their capacity to solve hard learning problems.
In the next section, you’ll look at the motivation behind an RNN.
Report an error