GraphDataLoaderBase<T>

Base Classes ยท AiDotNet.Data.Loaders

Abstract base class for graph data loaders providing common graph-related functionality.

Inheritance: DataLoaderBase<T> โ†’ GraphDataLoaderBase<T>

Implements: IGraphDataLoader<T>

Known derived types: CitationNetworkLoader<T>, MolecularDatasetLoader<T>, OGBDatasetLoader<T>

For Beginners

This base class handles common graph operations:

  • Storing node features and edge connections
  • Creating different types of tasks (node classification, link prediction)
  • Splitting data for training and evaluation

Concrete implementations (CitationNetworkLoader, MolecularDatasetLoader) extend this to load specific graph datasets.

How It Works

GraphDataLoaderBase provides shared implementation for all graph data loaders including:

  • Node feature and adjacency matrix management
  • Task creation (node classification, graph classification, link prediction)
  • Train/validation/test mask generation
  • Batch iteration for multiple graphs

Properties

Property Summary
Tensor<T> AdjacencyMatrix { get; }
Tensor<T> EdgeIndex { get; }
Tensor<T> GraphLabels { get; set; }
Tensor<T> NodeFeatures { get; }
Tensor<T> NodeLabels { get; }
bool HasNext { get; }
int BatchSize { get; set; }
int NumClasses { get; }
int NumEdges { get; }
int NumGraphs { get; }
int NumNodeFeatures { get; }
int NumNodes { get; }
int TotalCount { get; }

Methods

Method Summary
GraphClassificationTask<T> CreateGraphClassificationTask(double trainRatio, double valRatio, int? seed)
GraphData<T> GetNextBatch()
IAsyncEnumerable<GraphData<T>> GetBatchesAsync(int? batchSize, bool shuffle, bool dropLast, int? seed, int prefetchCount, CancellationToken cancellationToken)
IEnumerable<GraphData<T>> GetBatches(int? batchSize, bool shuffle, bool dropLast, int? seed)
LinkPredictionTask<T> CreateLinkPredictionTask(double trainRatio, double negativeRatio, int? seed)
NodeClassificationTask<T> CreateNodeClassificationTask(double trainRatio, double valRatio, int? seed)
bool TryGetNextBatch(GraphData<T> batch)