LayerGradCAMExplanation<T>

Models & Types · AiDotNet.Interpretability.Explainers

GradCAM explanation result.

Constructors

Constructor Summary
LayerGradCAMExplanation(Vector<T> input, Matrix<T> gradcamMap, Tensor<T> upsampledMap, int targetClass, T prediction, int layerHeight, int layerWidth, int[] inputShape) Initializes a new LayerGradCAM explanation.

Properties

Property Summary
Matrix<T> GradCAMMap { get; } Gets the GradCAM activation map.
T Prediction { get; } Gets the prediction score.
Tensor<T> UpsampledMap { get; } Gets the upsampled map (if available).
Vector<T> Input { get; } Gets the original input.
int LayerHeight { get; } Gets the layer height.
int LayerWidth { get; } Gets the layer width.
int TargetClass { get; } Gets the target class.
int[] InputShape { get; } Gets the input shape.

Methods

Method Summary
IEnumerable<ValueTuple<int, int, T>> GetTopRegions(int k) Gets top activated regions.
Matrix<T> GetNormalizedMap() Gets normalized GradCAM map (0-1 range).
string ToString() Returns string representation.