In this chapter, we illustrated the use of LSTM networks for developing a movie review sentiment classification model. One of the problems faced by recurrent neural networks that we used in the previous chapter is that it involves difficulty in capturing long-term dependency that may exist between two words/integers in a sequence of words or integers. Long Short-Term Memory (LSTM) networks are designed to artificially retain long-term memories that are important when dealing with long sentences or a long sequence of integers.

In the next chapter, we will continue to work with text data and explore the use of Convolutional Recurrent Neural Networks (CRNNs), which combine the benefits of Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) into a single network. We will illustrate the use of this type of network with the help of an interesting and publicly available text dataset, reuter_50_50.