CausalDiscoveryAlgorithmType
Enums · AiDotNet.Enums
Specifies the algorithm to use for causal structure learning (DAG discovery).
For Beginners
These algorithms figure out which variables cause which other variables by analyzing patterns in your data. Think of it like a detective figuring out cause-and-effect relationships. Different algorithms are like different detective methods — some test independence relationships, some optimize a score, and some use advanced math to find the best graph.
How It Works
Causal discovery algorithms learn the causal structure (a Directed Acyclic Graph or DAG) from observational data. Different algorithms make different assumptions about the data (linearity, Gaussianity, faithfulness) and use different strategies (constraint testing, score optimization, continuous optimization).
Fields
| Field | Summary |
|---|---|
CausalDiscoveryAlgorithmType ANM |
ANM — Additive Noise Model. |
CausalDiscoveryAlgorithmType AVICI |
AVICI — Amortized Variational Inference for Causal Discovery. |
CausalDiscoveryAlgorithmType AmortizedCD |
Amortized Causal Discovery — meta-learning approach to causal discovery. |
CausalDiscoveryAlgorithmType BCDNets |
BCD-Nets — Scalable variational Bayesian Causal Discovery. |
CausalDiscoveryAlgorithmType BOSS |
BOSS — Bayesian Optimal Structure Search. |
CausalDiscoveryAlgorithmType BayesDAG |
BayesDAG — Bayesian DAG learning with direct parameterization. |
CausalDiscoveryAlgorithmType CAM |
CAM — Causal Additive Models. |
CausalDiscoveryAlgorithmType CAMUV |
CAM-UV — Causal Additive Models with Unobserved Variables. |
CausalDiscoveryAlgorithmType CASTLE |
CASTLE — Causal Structure Learning. |
CausalDiscoveryAlgorithmType CCDr |
CCDr — Concave penalized Coordinate Descent with reparameterization. |
CausalDiscoveryAlgorithmType CCM |
CCM — Convergent Cross-Mapping for detecting causality in dynamical systems. |
CausalDiscoveryAlgorithmType CDNOD |
CD-NOD — Causal Discovery from Nonstationary/heterogeneous Data. |
CausalDiscoveryAlgorithmType CGNN |
CGNN — Causal Generative Neural Networks. |
CausalDiscoveryAlgorithmType CORL |
CORL — Causal Order learning via Reinforcement Learning. |
CausalDiscoveryAlgorithmType CPC |
CPC — Conservative PC. |
CausalDiscoveryAlgorithmType CausalVAE |
CausalVAE — Variational Autoencoder for causal representation learning. |
CausalDiscoveryAlgorithmType DAGGNN |
DAG-GNN — Graph Neural Network for DAG structure learning. |
CausalDiscoveryAlgorithmType DAGMALinear |
DAGMA Linear — log-determinant acyclicity constraint via M-matrices. |
CausalDiscoveryAlgorithmType DAGMANonlinear |
DAGMA Nonlinear — extends DAGMA with neural network function approximation. |
CausalDiscoveryAlgorithmType DECI |
DECI — Deep End-to-end Causal Inference. |
CausalDiscoveryAlgorithmType DYNOTEARS |
DYNOTEARS — Dynamic NOTEARS for time series structure learning. |
CausalDiscoveryAlgorithmType DiBS |
DiBS — Differentiable Bayesian Structure Learning. |
CausalDiscoveryAlgorithmType DirectLiNGAM |
DirectLiNGAM — Direct method for LiNGAM without ICA. |
CausalDiscoveryAlgorithmType ExactSearch |
Exact Search — dynamic programming for exact structure learning (exponential complexity). |
CausalDiscoveryAlgorithmType FCI |
FCI — Fast Causal Inference. |
CausalDiscoveryAlgorithmType FGES |
FGES — Fast Greedy Equivalence Search. |
CausalDiscoveryAlgorithmType FastIAMB |
Fast-IAMB — Faster variant of IAMB. |
CausalDiscoveryAlgorithmType GAE |
GAE — Graph Autoencoder for structure learning. |
CausalDiscoveryAlgorithmType GES |
GES — Greedy Equivalence Search. |
CausalDiscoveryAlgorithmType GFCI |
GFCI — Greedy FCI. |
CausalDiscoveryAlgorithmType GOBNILP |
GOBNILP — Integer Linear Programming for exact Bayesian network structure learning. |
CausalDiscoveryAlgorithmType GOLEM |
GOLEM — likelihood-based single-loop optimization without augmented Lagrangian. |
CausalDiscoveryAlgorithmType GRaSP |
GRaSP — Greedy Relaxation of the Sparsest Permutation. |
CausalDiscoveryAlgorithmType GraNDAG |
GraNDAG — Gradient-based Neural DAG Learning. |
CausalDiscoveryAlgorithmType GrangerCausality |
Granger Causality — tests whether one time series helps predict another. |
CausalDiscoveryAlgorithmType H2PC |
H2PC — Hybrid HPC algorithm. |
CausalDiscoveryAlgorithmType HillClimbing |
Hill Climbing — greedy local search with BIC or BDeu scoring. |
CausalDiscoveryAlgorithmType IAMB |
IAMB — Incremental Association Markov Blanket. |
CausalDiscoveryAlgorithmType ICALiNGAM |
ICA-LiNGAM — Linear Non-Gaussian Acyclic Model using Independent Component Analysis. |
CausalDiscoveryAlgorithmType IGCI |
IGCI — Information-Geometric Causal Inference. |
CausalDiscoveryAlgorithmType IterativeMCMC |
Iterative MCMC — Iterative Bayesian structure learning. |
CausalDiscoveryAlgorithmType K2 |
K2 Algorithm — score-based search with a known variable ordering. |
CausalDiscoveryAlgorithmType KraskovMI |
Kraskov Mutual Information — k-nearest neighbor mutual information estimator. |
CausalDiscoveryAlgorithmType LPCMCI |
LPCMCI — Latent PCMCI for time series with latent confounders. |
CausalDiscoveryAlgorithmType MCSL |
MCSL — Multi-scale Causal Structure Learning. |
CausalDiscoveryAlgorithmType MMHC |
MMHC — Max-Min Hill Climbing. |
CausalDiscoveryAlgorithmType MMPC |
MMPC — Max-Min Parents and Children. |
CausalDiscoveryAlgorithmType MarkovBlanket |
Markov Blanket discovery via the Grow-Shrink algorithm. |
CausalDiscoveryAlgorithmType NOTEARSLinear |
NOTEARS Linear — continuous optimization with tr(e^(W∘W))-d acyclicity constraint. |
CausalDiscoveryAlgorithmType NOTEARSLowRank |
NOTEARS Low-Rank — low-rank approximation for scalability to high dimensions. |
CausalDiscoveryAlgorithmType NOTEARSNonlinear |
NOTEARS Nonlinear — extends NOTEARS with MLP (multi-layer perceptron) for nonlinear relationships. |
CausalDiscoveryAlgorithmType NOTEARSSobolev |
NOTEARS with Sobolev basis functions for nonlinear relationships. |
CausalDiscoveryAlgorithmType NTSNOTEARS |
NTS-NOTEARS — Non-stationary Time Series NOTEARS. |
CausalDiscoveryAlgorithmType NeuralGranger |
Neural Granger Causality — deep learning extension of Granger causality. |
CausalDiscoveryAlgorithmType NoCurl |
NoCurl — curl-free constraint for acyclicity. |
CausalDiscoveryAlgorithmType OCSE |
oCSE — Optimal Causation Entropy for detecting causal relationships. |
CausalDiscoveryAlgorithmType OrderMCMC |
Order MCMC — MCMC over topological orderings for Bayesian structure learning. |
CausalDiscoveryAlgorithmType PC |
PC Algorithm — the gold standard constraint-based method using conditional independence tests. |
CausalDiscoveryAlgorithmType PCMCI |
PCMCI — PC algorithm adapted for time series with momentary conditional independence. |
CausalDiscoveryAlgorithmType PCMCIPlus |
PCMCI+ — Extension of PCMCI that also discovers contemporaneous causal links. |
CausalDiscoveryAlgorithmType PCNOTEARS |
PC-NOTEARS — Hybrid combining PC skeleton with NOTEARS optimization. |
CausalDiscoveryAlgorithmType PNL |
PNL — Post-Nonlinear causal model. |
CausalDiscoveryAlgorithmType PartitionMCMC |
Partition MCMC — MCMC over DAG partitions. |
CausalDiscoveryAlgorithmType RCD |
RCD — Repetitive Causal Discovery. |
CausalDiscoveryAlgorithmType RFCI |
RFCI — Really Fast Causal Inference. |
CausalDiscoveryAlgorithmType RSMAX2 |
RSMAX2 — Restricted maximization algorithm. |
CausalDiscoveryAlgorithmType TCDF |
TCDF — Temporal Causal Discovery Framework. |
CausalDiscoveryAlgorithmType TSFCI |
tsFCI — Time series Fast Causal Inference. |
CausalDiscoveryAlgorithmType TabuSearch |
Tabu Search — hill climbing with a tabu list to escape local optima. |
CausalDiscoveryAlgorithmType TiMINo |
TiMINo — Time series Model with Independent Noise. |
CausalDiscoveryAlgorithmType TransferEntropy |
Transfer Entropy — information-theoretic measure of directed information flow. |
CausalDiscoveryAlgorithmType VARLiNGAM |
VAR-LiNGAM — LiNGAM for time series via Vector Autoregressive model. |
int value__ |