Number of Input Nodes and Its States
Explore the impact of input nodes and their states on the structure and complexity of Bayesian networks. Understand how to balance model interpretability with computational demands by tuning these hyperparameters through iterative evaluation and expert collaboration. This lesson guides you in managing challenges like data sparsity, overfitting, and combinatorial explosion to build efficient and accurate models.
In this lesson, we cover the concept of input nodes. They represent the variables that serve as the starting points of the BN, which are often the observable variables in a dataset. The number of input nodes determines the initial complexity of the network. The input nodes, also known as root nodes or parentless nodes, are the nodes that have no incoming edges or connections from other nodes in the network. These nodes are typically used to model the initial or prior probabilities of certain events or variables.
Several networks for one problem
Let's say we want to model the different causes of lung cancer. Given different causes, and different possible ways of measuring these causes. For example, we can see whether a person is a smoker or not, but also we can add information on the frequency of smoking.
In the images below there are four different configurations to model this problem using a BN. Each one has a different number of input nodes and states of these nodes.
What is the size of the CDP table of the Lung Cancer ...