Struct LIME 

pub struct LIME { /* private fields */ }

LIME (Local Interpretable Model-agnostic Explanations).

Explains predictions by fitting a simple interpretable model locally around the instance being explained.

Algorithm

1. Generate perturbed samples around the instance
2. Get model predictions for each perturbation
3. Weight samples by proximity to original instance
4. Fit weighted linear model to the perturbations
5. Return linear model coefficients as explanations

Reference

• Ribeiro, M. T., et al. (2016). “Why Should I Trust You?”: Explaining the Predictions of Any Classifier. KDD.

Implementations

impl LIME

pub fn new(n_samples: usize, kernel_width: f32) -> Self

Create new LIME explainer.

Arguments

• n_samples - Number of perturbations to generate (default: 1000)
• kernel_width - Width of exponential kernel (default: 0.75)

pub fn explain<F>( &self, model_fn: F, sample: &Vector<f32>, seed: u64, ) -> LIMEExplanation

where F: Fn(&Vector<f32>) -> f32,

Explain a prediction by fitting a local linear model.

Arguments

• model_fn - Prediction function
• sample - Instance to explain
• seed - Random seed for reproducibility

Returns

Coefficients of the local linear model (feature importance)

pub fn n_samples(&self) -> usize

pub fn kernel_width(&self) -> f32

Trait Implementations

impl Debug for LIME

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for LIME

fn default() -> Self

Returns the “default value” for a type.