DeepCausalBase<T>
Base Classes · AiDotNet.CausalDiscovery.DeepLearning
Base class for deep learning-based causal discovery algorithms.
Inheritance: CausalDiscoveryBase<T> → DeepCausalBase<T>
Known derived types: AVICIAlgorithm<T>, AmortizedCDAlgorithm<T>, CASTLEAlgorithm<T>, CGNNAlgorithm<T>, CausalVAEAlgorithm<T>, DAGGNNAlgorithm<T>, DECIAlgorithm<T>, GAEAlgorithm<T>, GraNDAGAlgorithm<T>, TCDFAlgorithm<T>
For Beginners
These methods use neural networks to discover causal relationships. They can capture complex nonlinear effects but require more data and computation than traditional methods. Think of them as “letting the neural network figure out which variables cause which” by training it on the data.
How It Works
Deep learning methods learn causal structure by training neural networks that parameterize the structural equation model. The DAG constraint is typically enforced through continuous relaxation (e.g., NOTEARS-style) during training.
Properties
| Property | Summary |
|---|---|
CausalDiscoveryCategory Category { get; } |