DatasetSplit

Enums · AiDotNet.MetaLearning.Data

Represents the type of dataset split.

For Beginners

In machine learning, we typically split our data into three parts:

  • Train: Used during meta-training to learn how to learn
  • Validation: Used to tune hyperparameters without overfitting
  • Test: Used for final evaluation to see how well the model generalizes

In meta-learning, each split contains different classes to ensure the model learns to generalize to completely new tasks, not just new examples of seen classes.

Fields

Field Summary
DatasetSplit Test Test split used for final evaluation.
DatasetSplit Train Training split used for meta-training.
DatasetSplit Validation Validation split used for hyperparameter tuning and early stopping.
int value__