Conditional Probability Table (CPT)
Definition
A table that specifies the probability distribution of a node given every combination of states of its parent nodes. Every non-root node in a Bayesian network requires a CPT; root nodes require only a marginal probability distribution.
- Field
- Bayesian networks
- Content
- Probability distribution given parent node states
- Applies to
- Every non-root node
- Root nodes require
- Marginal probability distribution only
Common questions
Why does the size of a CPT grow so quickly as more parent nodes are added to a Bayesian network?+
A CPT must specify a probability for every combination of parent states, so the number of entries grows multiplicatively with each additional parent and with each parent's number of states, which is why network designers try to limit the number of parents per node and use structured or simplified relationships where possible.
How are CPT values typically populated in a forensic Bayesian network used to evaluate evidence?+
Values come from a mix of empirical data (such as measured error rates or population frequencies), expert elicitation where data is unavailable, and logical or deterministic relationships for nodes whose outcome follows necessarily from their parents' states. The source of each value is documented so the network's assumptions can be examined and challenged.
Related terms
- Belief Propagation
- An algorithm for computing marginal and posterior probabilities in a Bayesian network by passing messages between neighbouring nodes. Exact on tree-structured networks;...
- D-Separation
- A graphical criterion that determines whether two sets of nodes in a Bayesian network are conditionally independent given a third set. If...
- Directed Acyclic Graph (DAG)
- A graph in which edges have a direction (from parent to child node) and no path can return to a node it...
- Mixture Likelihood Ratio
- The ratio of the probability of observing a mixed DNA profile if the person of interest is a contributor to the probability...
- Sensitivity Analysis
- A technique for assessing how much the posterior probabilities in a Bayesian network change when the values in the conditional probability tables...