TVAEOptions<T>
Options & Configuration · AiDotNet.Models.Options
Configuration options for TVAE (Tabular Variational Autoencoder), a VAE-based model for generating realistic synthetic tabular data.
Inheritance: ModelOptions → NeuralNetworkOptions → ModelHyperparameterOptions → FinancialNeuralNetworkOptions → RiskModelOptions<T> → SyntheticDataGeneratorOptions<T> → TVAEOptions<T>
For Beginners
TVAE learns to compress your data into a small “summary” (latent space) and then reconstruct it back. Think of it like learning a recipe:
- Encoder: Looks at a data row and writes a compact “recipe” (latent code)
- Decoder: Reads the recipe and recreates the data row
- Training: The model learns to write good recipes that can recreate realistic data
- Generation: Sample random recipes from the latent space and decode them into new rows
TVAE is often faster to train than CTGAN and works well for moderate-sized datasets.
Example:
How It Works
TVAE uses a variational autoencoder architecture adapted for tabular data:
- Same VGM normalization and one-hot encoding as CTGAN for preprocessing
- Encoder compresses data into a latent Gaussian distribution (mean, logvar)
- Decoder reconstructs data from sampled latent codes via reparameterization trick
- ELBO loss balances reconstruction quality with KL divergence regularization
Reference: “Modeling Tabular Data using Conditional GAN” (Xu et al., NeurIPS 2019)
Constructors
| Constructor | Summary |
|---|---|
TVAEOptions() |
Initializes a new instance with the paper-default TVAE settings. |
TVAEOptions(TVAEOptions<T> other) |
Initializes a new instance by copying all user-configurable settings. |
Properties
| Property | Summary |
|---|---|
double LearningRate { get; set; } |
Gets or sets the learning rate for the optimizer. |
double LossWeight { get; set; } |
Gets or sets the weight for reconstruction loss relative to KL divergence. |
int BatchSize { get; set; } |
Gets or sets the training batch size. |
int Epochs { get; set; } |
Gets or sets the number of training epochs. |
int LatentDimension { get; set; } |
Gets or sets the dimension of the latent space. |
int VGMModes { get; set; } |
Gets or sets the number of Gaussian mixture components for VGM normalization. |
int[] DecoderDimensions { get; set; } |
Gets or sets the hidden layer sizes for the decoder network. |
int[] EncoderDimensions { get; set; } |
Gets or sets the hidden layer sizes for the encoder network. |