LIME

LIME: A technique that explains a prediction by locally approximating the model with an interpretable one. It generates perturbations around an example and fits a simple surrogate weighted by proximity. The result is local and can vary with sampling and representation.

In italiano: LIME Legge V · Overfitting, bias e collasso dei modelli

It generates perturbations around an example and fits a simple surrogate weighted by proximity. The result is local and can vary with sampling and representation.

Examples

Linear surrogate for text
Image superpixels
Explaining a single decision

Fonti

Why Should I Trust You? Explaining the Predictions of Any Classifier