FTTransformerRegression<T>

Models & Types · AiDotNet.NeuralNetworks.Tabular

FT-Transformer implementation for regression tasks.

Inheritance: FTTransformerBase<T> → FTTransformerRegression<T>

For Beginners

Use this model when you want to predict continuous numbers (like house prices, temperatures, stock returns, etc.).

How it works:

  1. Features are tokenized and processed by transformer layers
  2. The [CLS] token captures information from all features
  3. A linear layer maps the [CLS] representation to the output value(s)

Example:

How It Works

FTTransformerRegression applies the FT-Transformer architecture to regression problems. It uses the [CLS] token output with a linear regression head to predict continuous values.

Constructors

Constructor Summary
FTTransformerRegression(int numNumericalFeatures, int outputDimension, FTTransformerOptions<T> options) Initializes a new instance of the FTTransformerRegression class.

Properties

Property Summary
int OutputDimension { get; } Gets the output dimension (number of target values to predict).

Methods

Method Summary
T ComputeMAELoss(Tensor<T> predictions, Tensor<T> targets) Computes the Mean Absolute Error (MAE) loss.
T ComputeMAPE(Tensor<T> numericalFeatures, Tensor<T> targets, Matrix<int> categoricalIndices) Computes the Mean Absolute Percentage Error (MAPE).
T ComputeMSELoss(Tensor<T> predictions, Tensor<T> targets) Computes the Mean Squared Error (MSE) loss.
T ComputeR2Score(Tensor<T> numericalFeatures, Tensor<T> targets, Matrix<int> categoricalIndices) Computes the R² score (coefficient of determination).
T ComputeRMSE(Tensor<T> numericalFeatures, Tensor<T> targets, Matrix<int> categoricalIndices) Computes the Root Mean Squared Error (RMSE).
T TrainStep(Tensor<T> numericalFeatures, Tensor<T> targets, T learningRate, Matrix<int> categoricalIndices) Performs a single training step using MSE loss.
Tensor<T> Forward(Tensor<T> numericalFeatures) Performs the forward pass with numerical features only.
Tensor<T> Forward(Tensor<T> numericalFeatures, Matrix<int> categoricalIndices) Performs the forward pass to get predictions.
Tensor<T> Predict(Tensor<T> numericalFeatures) Alias for Forward with numerical features only.
Tensor<T> Predict(Tensor<T> numericalFeatures, Matrix<int> categoricalIndices) Alias for Forward - makes predictions on the input data.
void ResetState() Resets internal state.
void UpdateParameters(T learningRate) Updates all parameters including the regression head.