AI-ordlista
Den kompletta ordlistan över AI
Causal Deep Learning
Branch of deep learning that integrates principles of causal theory to discover and model cause-and-effect relationships in data, beyond simple correlations.
Propensity Score Matching
Technique where treated and untreated units are matched based on similar propensity scores to create a pseudo-randomized trial and estimate the causal effect.
Causal Neural Network
Neural network architecture explicitly designed to incorporate causal constraints or structures, to improve generalization and interpretability of predictions.
Causal Structure Discovery
Set of algorithms that aim to automatically learn the causal graph (cause-and-effect relationships) from observational data, often based on conditional independence tests.
Rubin Causality
Approach to causality based on the potential outcomes framework, where each unit has potential outcomes for each treatment state, of which only one is observed.
Potential Outcomes Model
Formalism where the causal effect is defined as the difference between potential outcomes under treatment and control for the same unit, foundation of Rubin causality.
Heterogeneous Treatment Effect (HTE)
Variation of the causal effect of an intervention across different subpopulations or individuals, which deep causal models seek to estimate accurately.
Instrumental Variable (IV)
Variable used to estimate a causal effect in the presence of unmeasured confounding, correlated with the treatment variable but not directly with the outcome, except through the treatment.