CausalDiscoveryCategory
Enums · AiDotNet.Enums
Categories of causal discovery algorithms based on their methodology.
For Beginners
Causal discovery algorithms can be grouped by how they work. Some test statistical independence, some optimize a score, some use continuous math, and some combine multiple approaches. This enum helps you understand and filter algorithms by their methodology.
Fields
| Field | Summary |
|---|---|
CausalDiscoveryCategory Bayesian |
Bayesian methods that maintain a posterior distribution over graph structures. |
CausalDiscoveryCategory ConstraintBased |
Constraint-based methods that use conditional independence tests to build the graph skeleton. |
CausalDiscoveryCategory ContinuousOptimization |
Continuous optimization methods that formulate DAG learning as a smooth optimization problem. |
CausalDiscoveryCategory DeepLearning |
Deep learning methods that use neural networks for structure learning. |
CausalDiscoveryCategory Functional |
Functional causal model methods that exploit properties of the noise distribution (e.g., non-Gaussianity). |
CausalDiscoveryCategory Hybrid |
Hybrid methods that combine constraint-based and score-based approaches. |
CausalDiscoveryCategory InformationTheoretic |
Information-theoretic methods that use entropy and mutual information measures. |
CausalDiscoveryCategory ScoreBasedSearch |
Score-based search methods that evaluate candidate DAGs using a scoring function (e.g., BIC, BDeu). |
CausalDiscoveryCategory Specialized |
Specialized methods that use unique mathematical formulations (e.g., integer linear programming). |
CausalDiscoveryCategory TimeSeries |
Time series causal discovery methods that account for temporal ordering and lagged effects. |
int value__ |