OcclusionExplanation<T>

Models & Types · AiDotNet.Interpretability.Explainers

Result of occlusion analysis.

Constructors

Constructor Summary
OcclusionExplanation(Tensor<T> input, Tensor<T> sensitivityMap, int targetClass, T basePrediction, int[] windowShape, int[] strides) Initializes a new occlusion explanation.

Properties

Property Summary
T BasePrediction { get; } Gets the base prediction for the target class.
Tensor<T> Input { get; } Gets the original input.
Tensor<T> SensitivityMap { get; } Gets the sensitivity map showing importance of each region.
int TargetClass { get; } Gets the target class that was explained.
int[] Strides { get; } Gets the strides used for sliding.
int[] WindowShape { get; } Gets the window shape used for occlusion.

Methods

Method Summary
IEnumerable<ValueTuple<int[], T>> GetMostImportantRegions(int k) Gets the most important regions (highest sensitivity).
Tensor<T> GetUpsampledMap() Upsamples the sensitivity map to match input size.
string ToString() Returns a human-readable summary.