IGPUAcceleratedExplainer<T>

Interfaces · AiDotNet.Interpretability.Helpers

Interface for explainers that support GPU acceleration.

Known implementations: DeepLIFTExplainer<T>, DeepSHAPExplainer<T>, DiCEExplainer<T>, FeatureAblationExplainer<T>, GradientSHAPExplainer<T>, GuidedBackpropExplainer<T>, GuidedGradCAMExplainer<T>, InfluenceFunctionExplainer<T>, InputXGradientExplainer<T>, IntegratedGradientsExplainer<T>, LIMEExplainer<T>, LayerAttributionExplainer<T>, LayerGradCAMExplainer<T>, NeuronAttributionExplainer<T>, OcclusionExplainer<T>, PermutationFeatureImportance<T>, SHAPExplainer<T>, TCAVExplainer<T>, TreeSHAPExplainer<T>

For Beginners

Explainers implementing this interface can leverage GPU hardware for faster computation. The GPU helper is optional - if not provided, the explainer falls back to CPU computation.

To enable GPU acceleration:

  1. Create a GPUExplainerHelper with your GPU runtime
  2. Pass it to the explainer via SetGPUHelper()
  3. Explanations will automatically use GPU when beneficial

Properties

Property Summary
bool IsGPUAccelerated { get; } Gets whether GPU acceleration is currently enabled.

Methods

Method Summary
void SetGPUHelper(GPUExplainerHelper<T> helper) Sets the GPU helper for accelerated computation.