Enum DatasetSplit

Namespace

AiDotNet.MetaLearning.Data

Assembly

AiDotNet.dll

Represents the type of dataset split.

public enum DatasetSplit

Fields

Test = 2

Test split used for final evaluation.

Train = 0

Training split used for meta-training.

Validation = 1

Validation split used for hyperparameter tuning and early stopping.

Remarks

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.