SvmSmoteAugmenter<T>

Models & Types · AiDotNet.Augmentation.Tabular

Implements SVM-SMOTE for imbalanced datasets, using SVM decision boundary to identify borderline samples.

Inheritance: AugmentationBase<T, Matrix<T>> → TabularAugmenterBase<T> → SvmSmoteAugmenter<T>

For Beginners

SVM-SMOTE is an enhancement over standard SMOTE that uses Support Vector Machine classification to identify the most informative samples near the decision boundary. It generates synthetic samples from support vectors (the most difficult cases), focusing oversampling effort where it matters most.

How It Works

How it works:

  • Train an SVM classifier on the data to identify support vectors
  • Identify minority samples that are support vectors (near the decision boundary)
  • Generate synthetic samples by interpolating between support vector minority samples and their neighbors

When to use:

  • When the decision boundary is critical for classification
  • When you want synthetic samples focused on difficult cases
  • When standard SMOTE generates too many samples in easy regions

Reference: Nguyen et al., “Borderline Over-Sampling for Imbalanced Data Classification” (2011)

Constructors

Constructor Summary
SvmSmoteAugmenter(int kNeighbors, double samplingRatio, double svmC, int maxIterations, double probability) Creates a new SVM-SMOTE augmenter.

Properties

Property Summary
double SamplingRatio { get; } Gets the sampling ratio for synthetic sample generation.
double SvmC { get; } Gets the SVM regularization parameter (C).
int KNeighbors { get; } Gets the number of nearest neighbors to use for interpolation.
int MaxIterations { get; } Gets the maximum iterations for SVM training.

Methods

Method Summary
IDictionary<string, object> GetParameters()
Matrix<T> GenerateSyntheticSamples(Matrix<T> minorityData, Matrix<T> majorityData, AugmentationContext<T> context) Applies SVM-SMOTE to generate synthetic samples for the minority class.