GOGGLEOptions<T>
Options & Configuration · AiDotNet.Models.Options
Configuration options for GOGGLE (Generative mOdelling for tabular data by learninG reLational structurE), a graph-based VAE that learns feature dependency structure for generating realistic tabular data.
Inheritance: ModelOptions → NeuralNetworkOptions → ModelHyperparameterOptions → FinancialNeuralNetworkOptions → RiskModelOptions<T> → SyntheticDataGeneratorOptions<T> → GOGGLEOptions<T>
For Beginners
GOGGLE discovers which features depend on each other and uses this knowledge to generate better data:
- Learns a “dependency map” — e.g., “Income depends on Education and Age”
- Features that are connected share information through graph neural networks
- This produces data where feature relationships are more realistic
Example:
How It Works
GOGGLE combines a VAE with a graph neural network (GNN) to learn and exploit feature dependencies:
- Structure learning: Learns a soft adjacency matrix representing feature dependencies
- GNN encoder: Uses message passing on the learned graph to encode features
- VAE framework: Generates data through a latent space with reparameterization
- Structure loss: Regularizes the learned graph for sparsity and DAG properties
Reference: “GOGGLE: Generative Modelling for Tabular Data by Learning Relational Structure” (Liu et al., ICLR 2023)
Constructors
| Constructor | Summary |
|---|---|
GOGGLEOptions() |
Properties
| Property | Summary |
|---|---|
double KLWeight { get; set; } |
Gets or sets the weight for the KL divergence loss term. |
double LearningRate { get; set; } |
Gets or sets the learning rate. |
double SparsityWeight { get; set; } |
Gets or sets the sparsity regularization for the adjacency matrix. |
double StructureWeight { get; set; } |
Gets or sets the weight for the graph structure learning loss. |
int BatchSize { get; set; } |
Gets or sets the training batch size. |
int Epochs { get; set; } |
Gets or sets the number of training epochs. |
int HiddenDimension { get; set; } |
Gets or sets the hidden dimension for GNN and MLP layers. |
int LatentDimension { get; set; } |
Gets or sets the dimension of the VAE latent space. |
int NumGNNLayers { get; set; } |
Gets or sets the number of GNN message-passing layers. |
int VGMModes { get; set; } |
Gets or sets the number of VGM modes for continuous column transformation. |