LIMEExplainer<T>
Models & Types · AiDotNet.Interpretability.Explainers
Model-agnostic LIME (Local Interpretable Model-agnostic Explanations) explainer.
Implements: IGPUAcceleratedExplainer<T>, ILocalExplainer<T, LIMEExplanationResult<T>>
For Beginners
LIME explains individual predictions by building a simple, interpretable model (like linear regression) that approximates the complex model locally around the prediction you want to understand.
How it works:
- Generate perturbed samples near your instance (slightly modified versions)
- Get the complex model’s predictions for these samples
- Fit a simple linear model to these nearby predictions
- The linear model’s coefficients show which features matter most
Think of it like zooming in on a curvy road - if you zoom in enough, even a curved road looks straight. LIME zooms in on your prediction and fits a “straight line” (linear model) to explain it.
Constructors
| Constructor | Summary |
|---|---|
LIMEExplainer(Func<Matrix<T>, Vector<T>> predictFunction, int numFeatures, int nSamples, double kernelWidth, string[] featureNames, double[] featureStdDevs, int? randomState) |
Initializes a new LIME explainer. |
Properties
| Property | Summary |
|---|---|
bool IsGPUAccelerated { get; } |
|
bool SupportsGlobalExplanations { get; } |
|
bool SupportsLocalExplanations { get; } |
|
int NumFeatures { get; } |
Gets the number of features being explained. |
string MethodName { get; } |
Methods
| Method | Summary |
|---|---|
LIMEExplainer<T> FromModel(IFullModel<T, TInput, TOutput> model, int numFeatures, int nSamples, double kernelWidth, string[] featureNames, double[] featureStdDevs, int? randomState) |
Creates a LIME explainer from a model. |
LIMEExplanationResult<T> Explain(Vector<T> instance) |
|
LIMEExplanationResult<T>[] ExplainBatch(Matrix<T> instances) |
|
void SetGPUHelper(GPUExplainerHelper<T> helper) |