Interface IShuffleable

Namespace

AiDotNet.Interfaces

Assembly

AiDotNet.dll

Defines capability to shuffle data for randomized iteration.

public interface IShuffleable

Remarks

Data loaders that implement this interface can shuffle their data, which is important for training to prevent the model from learning the order of examples rather than the patterns in the data.

For Beginners: Shuffling is like shuffling a deck of cards before dealing. When training, you don't want your model to learn "cat images always come first, then dog images" - you want it to learn actual features. Shuffling ensures each epoch sees data in a different order.

Properties

IsShuffled

Gets whether the data is currently shuffled.

bool IsShuffled { get; }

Property Value

bool

Methods

Shuffle(int?)

Shuffles the data order using the specified seed for reproducibility.

void Shuffle(int? seed = null)

Parameters

seed int?

Optional seed for reproducible shuffling. Same seed produces same order.

Remarks

For Beginners: The seed is like a "recipe" for the randomness. Using the same seed gives the same "random" order every time, which is useful for reproducing experiments or debugging.

Unshuffle()

Restores the original (unshuffled) data order.

void Unshuffle()