3 Deep learning in a nutshell

 

This chapter covers

  • The basics of building and training deep learning models
  • Using a multilayer perceptron for regression on tabular data
  • Classifying image data with a multilayer perceptron and a convolutional neural network
  • Classifying text data with a recurrent neural network

Deep learning, a subfield of ML, has become a scorching topic in the AI community and beyond. It drives numerous applications across various fields and has achieved superior performance compared with many of the more traditional models introduced earlier. This chapter will present the basic building blocks of deep learning and show you how to apply three popular types of models to solve supervised learning tasks on different data types. The chapter will also serve as a stepping stone to help you better understand the AutoML methods for generating and tuning deep learning methods introduced in the second part of the book.

3.1 What is deep learning?

3.2 TensorFlow and Keras

3.3 California housing price prediction with a multilayer perceptron

3.3.1 Assembling and preparing the data

3.3.2 Building up the multilayer perceptron

3.3.3 Training and testing the neural network

3.3.4 Tuning the number of epochs

3.4 Classifying handwritten digits with convolutional neural networks

3.4.1 Assembling and preparing the dataset

3.4.2 Addressing the problem with an MLP

3.4.3 Addressing the problem with a CNN

3.5 IMDB review classification with recurrent neural networks

3.5.1 Preparing the data

3.5.2 Building up the RNN

3.5.3 Training and validating the RNN

Summary