Recurrent Neural Networks

In this section, we are going to review Recurrent Neural Networks (RNNs). This topic will first look at the theory of RNNs. It will review many architectures within this model and help you to work out which model to use to solve a certain problem, and it will also look at several types of RNN and their pros and cons. Also, we will look at how to create a simple RNN, train it, and make predictions.

Introduction to Recurrent Neural Networks (RNN)

Human behavior shows a variety of serially ordered action sequences. A human is capable of learning dynamic paths based on a set of previous actions or sequences. This means that people do not start learning from scratch; we have some previous knowledge, which helps us. For example you could not understand a word if you did not understand the previous word in a sentence!

Traditionally, neural networks cannot solve these types of problem because they cannot learn previous information. But what happens with problems that cannot...