GANDALFOptions<T>
Options & Configuration · AiDotNet.Models.Options
Configuration options for GANDALF (Gated Additive Neural Decision Forest).
Inheritance: ModelOptions → NeuralNetworkOptions → ModelHyperparameterOptions → FinancialNeuralNetworkOptions → RiskModelOptions<T> → GANDALFOptions<T>
For Beginners
GANDALF is like having a smart feature selector combined with a forest of decision trees.
Key ideas:
- Gated Feature Selection: Learns which features matter for each prediction
- Soft Decision Trees: Trees with smooth (differentiable) decisions
- Additive Ensemble: Trees are combined by adding their predictions
Why this works well:
- Automatic feature importance learning (no manual selection needed)
- Interpretable structure (can see which features and paths are used)
- Combines benefits of neural networks and decision trees
- Good handling of both numerical and categorical features
Example:
How It Works
GANDALF combines gated feature selection with neural decision trees in an additive ensemble. It learns which features are important through attention-based gating and makes predictions through soft decision trees.
Reference: “GANDALF: Gated Adaptive Network for Deep Automated Learning of Features” (2022)
Constructors
| Constructor | Summary |
|---|---|
GANDALFOptions() |
Properties
| Property | Summary |
|---|---|
bool UseBatchNorm { get; set; } |
Gets or sets whether to use batch normalization. |
double DropoutRate { get; set; } |
Gets or sets the dropout rate for regularization. |
double InitScale { get; set; } |
Gets or sets the initialization scale for tree parameters. |
double Temperature { get; set; } |
Gets or sets the temperature for soft tree decisions. |
double WeightDecay { get; set; } |
Gets or sets the weight decay for regularization. |
int GatingHiddenDimension { get; set; } |
Gets or sets the hidden dimension for gating network. |
int LeafDimension { get; set; } |
Gets or sets the leaf dimension (output dimension per leaf). |
int NumGatingLayers { get; set; } |
Gets or sets the number of gating layers. |
int NumInternalNodes { get; } |
Gets the number of internal nodes per tree (2^depth - 1). |
int NumLeaves { get; } |
Gets the number of leaves per tree (2^depth). |
int NumTrees { get; set; } |
Gets or sets the number of trees in the ensemble. |
int TreeDepth { get; set; } |
Gets or sets the depth of each tree. |