Struct GradientExplainer 

pub struct GradientExplainer<M> { /* private fields */ }

Expected Gradients (Gradient SHAP) with background interpolation and optional Gaussian local smoothing.

Implementations

impl<M> GradientExplainer<M>

pub fn new(model: M, background: Background) -> Self

pub fn with_nsamples(self, n: usize) -> Self

pub fn with_seed(self, s: u64) -> Self

pub fn with_local_smoothing(self, s: f64) -> Self

pub fn with_batch_size(self, batch_size: usize) -> Self

Limits gradient rows submitted to the autodiff backend in one call.

impl<M: DifferentiablePredict> GradientExplainer<M>

pub fn explain_with_uncertainty( &self, x: ArrayView2<'_, f64>, repeats: usize, ) -> Result<UncertainExplanation>

Repeats Expected Gradients with independent deterministic seeds and returns the standard error of the mean attribution.

Trait Implementations

impl<M: DifferentiablePredict> Explainer for GradientExplainer<M>

fn explain(&self, x: ArrayView2<'_, f64>) -> Result<Explanation>