AutoDiffTabGenerator<T>
Models & Types · AiDotNet.NeuralNetworks.SyntheticData
AutoDiff-Tab generator that automatically searches over diffusion configurations (timesteps, noise schedules, network architecture) to find optimal settings for tabular data generation.
Inheritance: NeuralNetworkBase<T> → NeuralSyntheticTabularGeneratorBase<T> → AutoDiffTabGenerator<T>
Implements: IGeneratedParameterRegistrar<T>, ISyntheticTabularGenerator<T>
For Beginners
AutoDiff-Tab is a “smart” version of TabDDPM:
- It tries several different diffusion setups (like testing different recipes)
- Evaluates which one works best on your data
- Uses the winner for the final model
This saves you from having to manually tune hyperparameters.
This implementation follows the standard neural network architecture pattern with:
- Proper inheritance from NeuralNetworkBase
- Layer-based architecture using ILayer components
- Engine-based tensor operations for CPU/GPU acceleration
- Full autodiff and JIT compilation support
If you provide custom layers in the architecture, those will be used directly for the denoiser network. Otherwise, the network creates the standard architecture based on the best configuration found during search.
Example usage:
How It Works
AutoDiff-Tab combines architecture search with diffusion models:
Reference: “Automated Diffusion Models for Tabular Data” (2024)
Constructors
| Constructor | Summary |
|---|---|
AutoDiffTabGenerator() |
Initializes a new AutoDiff-Tab generator with the specified architecture. |
AutoDiffTabGenerator(NeuralNetworkArchitecture<T> architecture, AutoDiffTabOptions<T> options, IGradientBasedOptimizer<T, Tensor<T>, Tensor<T>> optimizer, ILossFunction<T> lossFunction) |
Initializes a new instance of the AutoDiffTabGenerator. |
Properties
| Property | Summary |
|---|---|
AutoDiffTabOptions<T> Options { get; } |
Gets the AutoDiffTab-specific options. |
IReadOnlyList<ColumnMetadata> Columns { get; } |
|
bool IsFitted { get; set; } |
Methods
| Method | Summary |
|---|---|
Dictionary<string, T> GetFeatureImportance() |
|
Matrix<T> Generate(int numSamples, Vector<T> conditionColumn, Vector<T> conditionValue) |
Generates new synthetic tabular data rows using the trained diffusion model. |
ModelMetadata<T> GetModelMetadata() |
|
Task FitAsync(Matrix<T> data, IReadOnlyList<ColumnMetadata> columns, int? epochs, CancellationToken ct) |
|
Tensor<T> ForwardForTraining(Tensor<T> input) |
|
void Fit(Matrix<T> data, IReadOnlyList<ColumnMetadata> columns, int? epochs) |
Fits the AutoDiff-Tab generator to the provided real tabular data. |