AI 용어집
인공지능 완전 사전
Confounder
Variable that influences both the exposure and the outcome, creating a spurious association that must be controlled to correctly estimate the causal effect between these two variables.
Backdoor Criterion
Set of rules allowing the identification of variables to adjust for in order to block all non-causal paths between treatment and outcome, while preserving causal paths in a DAG.
Frontdoor Criterion
Alternative identification condition to the backdoor criterion, allowing the estimation of causal effects through mediators even in the presence of unmeasured confounders affecting the treatment and outcome.
Markov Blanket
Minimal set of variables that make a node independent of all other nodes in the graph conditionally, consisting of its parents, children, and the parents of its children.
Adjustment Set
Set of variables to control for in order to isolate the causal effect between two variables, identified by graphical criteria such as the backdoor criterion to ensure unbiased estimation.
Causal Paths
Sequences of directed edges in a DAG representing the mechanisms by which one variable influences another, essential for the identification and interpretation of direct and indirect causal effects.