{"paper":{"title":"Scalable unsupervised feature selection via weight stability","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"Relevant features are assigned consistently higher weights than noise features across a range of Minkowski exponents.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Renato Cordeiro de Amorim, Xudong Zhang","submitted_at":"2025-06-06T14:24:41Z","abstract_excerpt":"Unsupervised feature selection is critical for improving clustering performance in high-dimensional data, where irrelevant features can obscure meaningful structure. In this work, we introduce the Minkowski weighted $k$-means++, a novel initialisation strategy for the Minkowski Weighted $k$-means. Our initialisation selects centroids probabilistically using feature relevance estimates derived from the data itself. Building on this, we propose two new feature selection algorithms, FS-MWK++, which aggregates feature weights across a range of Minkowski exponents to identify stable and informative"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Under explicit assumptions on noise features and cluster structure, relevant features are assigned consistently higher weights than noise features across a range of Minkowski exponents.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The paper relies on explicit assumptions about the properties of noise features and the underlying cluster structure that allow the weight stability to separate signal from noise.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Proposes FS-MWK++ and scalable SFS-MWK++ for unsupervised feature selection by aggregating stable feature weights from Minkowski weighted k-means++ across exponents, backed by theoretical analysis under noise and cluster assumptions.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Relevant features are assigned consistently higher weights than noise features across a range of Minkowski exponents.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"75ebb09882505900bddc2665017a21a72536cb8db59be01d5cf35c3748f1efd0"},"source":{"id":"2506.06114","kind":"arxiv","version":5},"verdict":{"id":"9761e992-79d1-49ac-ada3-a7666dae5802","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T10:26:36.878416Z","strongest_claim":"Under explicit assumptions on noise features and cluster structure, relevant features are assigned consistently higher weights than noise features across a range of Minkowski exponents.","one_line_summary":"Proposes FS-MWK++ and scalable SFS-MWK++ for unsupervised feature selection by aggregating stable feature weights from Minkowski weighted k-means++ across exponents, backed by theoretical analysis under noise and cluster assumptions.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The paper relies on explicit assumptions about the properties of noise features and the underlying cluster structure that allow the weight stability to separate signal from noise.","pith_extraction_headline":"Relevant features are assigned consistently higher weights than noise features across a range of Minkowski exponents."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2506.06114/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"5dd76052a0f0cdfcd3d284a62af23465ae65f875659988cb4b5778cf68b1c389"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}