2 Deep learning and language: The basics

 

This chapter covers

  • Highlighting the fundamental architectures of deep learning: multilayer perceptrons and spatial and temporal filtering
  • Introducing deep learning models for natural language processing

After reading this chapter, you will have a clear idea of how deep learning works in general, why it is different from other machine learning approaches, and what it brings to the field of natural language processing. This chapter introduces you to some Keras concepts and their implementation details through examples.

2.1 Basic architectures of deep learning

Let’s investigate the basic architectures of deep learning: multilayer perceptrons and different forms of filtering input, spatial and temporal (see figure 2.1).

Figure 2.1 Basic architectures of deep learning

2.1.1 Deep multilayer perceptrons

The prototypical deep learning network is a multilayer perceptron (MLP). We came across these in chapter 1, and we repeat the architecture in figure 2.2 for a simple multilayer perceptron that has only a single hidden layer.

2.1.2 Two basic operators: Spatial and temporal

2.2 Deep learning and NLP: A new paradigm

Summary