Home / Curated / Algorithms / Submodular pick LIME

Submodular pick LIME

Algorithms AIMech:SPLIME DRAFT

Aliases: SP-LIME

A method selecting a budgeted set of local LIME explanations by greedily increasing weighted coverage of explanatory features.

Identity Notes & Concept Scope SP-LIME selects representative local explanations; it is separate from fitting each LIME surrogate and does not produce a globally faithful surrogate model.

Claims & Grounded Evidence

1 Claim

Mechanism #explanation-method

Build an instance-by-feature explanation matrix and global feature weights, then select instances by marginal gain in the total weight of features covered by at least one selected explanation.

supports Source: Why Should I Trust You?: Explaining the Predictions of Any Classifier · Locator: Section 4, Algorithm 2 and equations 3–4, PDF/printed pp. 5–6

⚠️ Curated Limitations & Negative Results

Coverage depends on the candidate instances, local explanations, interpretable features and importance weights.
The greedy coverage objective is not a guarantee of overall model trustworthiness; cross-image feature comparability requires additional choices.

Artificial Intelligence Ontology (AIO) Grounding

Mapped AIO Classes

AIO Target Class	Mapping Relation	Mapping Rationale
`AIO:Model` Model	related_match	The procedure selects explanations of predictive-model behavior for inspection. This direct analysis relationship does not make the selector a model class.

Evidence Sources & Primary Literature

1 Source

Source ID	Publication / Document	Type / Access	Reference & Checksum
`original-paper`	Why Should I Trust You?: Explaining the Predictions of Any Classifier Marco Tulio Ribeiro, Sameer Singh, Carlos Guestrin	primary paper full_text	ID: arXiv:1602.04938v3 sha256:365079753571…

Curation Provenance

Curator: Codex

Method: ai_assisted

Model: GPT-6

Created: 2026-10-11T04:02:04Z

Updated: 2026-10-11T04:02:04Z

Source File: data/curated/SPLIME.yaml