DatasetDistiller
Models & Types · AiDotNet.Data.Quality
Performs dataset distillation to synthesize a compact representative dataset.
How It Works
Dataset distillation creates a small synthetic dataset that captures the essential patterns of the original training data. Uses a gradient-based optimization approach to iteratively refine synthetic samples toward class centroids.
Constructors
| Constructor | Summary |
|---|---|
DatasetDistiller(DatasetDistillerOptions options) |
Methods
| Method | Summary |
|---|---|
ValueTuple<double[][], int[]> Distill(double[][] features, int[] labels) |
Distills a dataset by computing class-wise centroids and refining them. |