5 Connecting causality and deep learning

 

This chapter covers

  • Incorporating deep learning into a causal graphical model
  • Training a causal graphical model with a variational autoencoder
  • Using causal methods to enhance machine learning

The title of this book is Causal AI, but how exactly does causality connect to AI? More specifically, how does causality connect with deep learning, the dominant paradigm in AI? In this chapter, I look at this question from two perspectives:

  • How to incorporate deep learning into a causal model—We’ll look at a causal model of a computer vision problem (section 5.1) and then train the deep causal image model (section 5.2).
  • How to use causal reasoning to do better deep learning—We’ll look at a case study on independence of mechanism and semi-supervised learning (section 5.3.1 and 5.3.2), and we’ll demystify deep learning with causality (section 5.3.3).

5.1 A causal model of a computer vision problem

5.1.1 Leveraging the universal function approximator

5.1.2 Causal abstraction and plate models

5.2 Training a neural causal model

5.2.1 Setting up the training data

5.2.2 Setting up the variational autoencoder

5.2.3 The training procedure

5.2.4 Evaluating training

5.2.5 How should we causally interpret Z?

5.2.6 Advantages of this causal interpretation

5.3 Using causal inference to enhance deep learning

5.3.1 Independence of mechanism as an inductive bias

5.3.2 Case study: Semi-supervised learning

5.3.3 Demystifying deep learning with causality

Summary