ABSTRACT
Conditional independence is a crucial concept in the network models discussed in this book, because the canonical pairwise Markov random field (PMRF) formulation of networks is defined in terms of this concept: nodes that are disconnected in the true PMRF are conditionally independent given the other nodes in the network. This chapter provides an introduction to the concept of conditional independence and illustrates its importance in reasoning about possible generative models. With the intuitive understanding of conditional independence provided in this chapter, readers will be able to grasp the essence of the estimation routines discussed in subsequent chapters.
