IFeatureAttribution<T>

Interfaces · AiDotNet.Interpretability

A common accessor for the per-feature attribution vector an explanation carries, regardless of the method that produced it (SHAP values, integrated gradients, DeepLIFT contributions, ablation deltas, saliency, …).

Known implementations: DeepLIFTExplanation<T>, DeepSHAPExplanation<T>, FeatureAblationExplanation<T>, GradientSHAPExplanation<T>, InputXGradientExplanation<T>, IntegratedGradientsExplanation<T>, SHAPExplanation<T>, SaliencyMapExplanation<T>, TreeSHAPExplanation<T>

How It Works

Each explanation type names its attributions differently (ShapValues, Attributions, Saliency, …). Implementing this lets downstream logic — most importantly the faithfulness audit — read attributions uniformly without knowing the concrete explanation type.

Methods

Method Summary
Vector<T> GetFeatureAttributions() Returns the per-feature attribution vector (one score per input feature).