SAINTBase<T>
Base Classes · AiDotNet.NeuralNetworks.Tabular
Base class for SAINT (Self-Attention and Intersample Attention Transformer).
Implements: IParameterSource<T>
Known derived types: SAINTClassifier<T>, SAINTRegression<T>
For Beginners
Think of SAINT as looking at your data from two perspectives:
- Column attention: “Which features are related to each other?”
(e.g., income and education level might be correlated)
- Row attention: “Which samples in my batch are similar?”
(e.g., customers with similar profiles might have similar behavior)
By combining both views, SAINT can learn patterns that other models miss.
How It Works
SAINT applies two types of attention in alternating layers:
- Column attention: Self-attention over features (like FT-Transformer)
- Row attention: Inter-sample attention comparing samples in a batch
Properties
| Property | Summary |
|---|---|
long ParameterCount { get; } |
Methods
| Method | Summary |
|---|---|
Vector<T> GetParameters() |
|
void ResetState() |
Resets internal state. |
void SetParameters(Vector<T> parameters) |
|
void UpdateParameters(T learningRate) |
Updates all parameters. |