{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:7ECUX2WT3FVS2IHK4OZBKP6AL6","short_pith_number":"pith:7ECUX2WT","schema_version":"1.0","canonical_sha256":"f9054bead3d96b2d20eae3b2153fc05f965498ef02a3c8ce2306eff1901e3ba4","source":{"kind":"arxiv","id":"2604.23354","version":2},"attestation_state":"computed","paper":{"title":"Explainable AI in Speaker Recognition -- Making Latent Representations Understandable","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Speaker recognition neural networks organize their latent representations into hierarchical clusters that align with semantic attributes like gender and nationality.","cross_cats":["cs.AI","eess.SP"],"primary_cat":"eess.AS","authors_text":"Mark D. Plumbley, Wenwu Wang, Yanze Xu","submitted_at":"2026-04-25T15:44:20Z","abstract_excerpt":"Neural networks can be trained to learn task-relevant representations from data. Understanding how these networks make decisions falls within the Explainable AI (XAI) domain. This paper proposes to study an XAI topic: uncovering the unknown organisation in the representations, particularly those a speaker recognition network learns from utterances, for recognising speaker identity.\n  Past studies have employed algorithms (e.g. K-means) to analyse how network representations can be naturally organised into independent clusters in different ways, i.e., to analyse flat clustering phenomena within"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2604.23354","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"eess.AS","submitted_at":"2026-04-25T15:44:20Z","cross_cats_sorted":["cs.AI","eess.SP"],"title_canon_sha256":"ae83f33cba4c15379fe40b1661e10e8ff8fdb173de9ca753b34d66df75b37c45","abstract_canon_sha256":"b846fd5324a4735d8795f09020c794678597c233ede5167750785e06b97572d1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-29T01:05:10.515376Z","signature_b64":"2xQg1PIfYOamGNIaSsEn0wazw9p62TTHyVKBA0LufUyICub+fpnvk8ujlKC95hCl3JLoizLu+NBMBIuxDf6hDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f9054bead3d96b2d20eae3b2153fc05f965498ef02a3c8ce2306eff1901e3ba4","last_reissued_at":"2026-05-29T01:05:10.514741Z","signature_status":"signed_v1","first_computed_at":"2026-05-29T01:05:10.514741Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Explainable AI in Speaker Recognition -- Making Latent Representations Understandable","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Speaker recognition neural networks organize their latent representations into hierarchical clusters that align with semantic attributes like gender and nationality.","cross_cats":["cs.AI","eess.SP"],"primary_cat":"eess.AS","authors_text":"Mark D. Plumbley, Wenwu Wang, Yanze Xu","submitted_at":"2026-04-25T15:44:20Z","abstract_excerpt":"Neural networks can be trained to learn task-relevant representations from data. Understanding how these networks make decisions falls within the Explainable AI (XAI) domain. This paper proposes to study an XAI topic: uncovering the unknown organisation in the representations, particularly those a speaker recognition network learns from utterances, for recognising speaker identity.\n  Past studies have employed algorithms (e.g. K-means) to analyse how network representations can be naturally organised into independent clusters in different ways, i.e., to analyse flat clustering phenomena within"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"This work applies SLINK and HDBSCAN to demonstrate the existence of hierarchical clustering phenomena within the network representation space, and designs HCCM to perform one-to-one matching between predefined semantic classes and hierarchical representation clusters, with Liebig's score to quantify performance.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the hierarchical clusters produced by SLINK or HDBSCAN correspond to meaningful semantic classes or their conjunctions in a non-arbitrary way that HCCM can reliably detect and that Liebig's score meaningfully diagnoses limiting factors.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Speaker recognition networks form hierarchical clusters in latent space that can be matched to semantic classes using new HCCM algorithm and quantified by Liebig's score.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Speaker recognition neural networks organize their latent representations into hierarchical clusters that align with semantic attributes like gender and nationality.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"95ac01f5a4f2f342730ed2bffdd8e2c841b50b8c74678370b22c7286c4506168"},"source":{"id":"2604.23354","kind":"arxiv","version":2},"verdict":{"id":"d070c761-f6c7-4423-95bd-7990dfc25d6d","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-08T06:54:00.456162Z","strongest_claim":"This work applies SLINK and HDBSCAN to demonstrate the existence of hierarchical clustering phenomena within the network representation space, and designs HCCM to perform one-to-one matching between predefined semantic classes and hierarchical representation clusters, with Liebig's score to quantify performance.","one_line_summary":"Speaker recognition networks form hierarchical clusters in latent space that can be matched to semantic classes using new HCCM algorithm and quantified by Liebig's score.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the hierarchical clusters produced by SLINK or HDBSCAN correspond to meaningful semantic classes or their conjunctions in a non-arbitrary way that HCCM can reliably detect and that Liebig's score meaningfully diagnoses limiting factors.","pith_extraction_headline":"Speaker recognition neural networks organize their latent representations into hierarchical clusters that align with semantic attributes like gender and nationality."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.23354/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-21T09:34:00.617979Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T23:14:59.591565Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"9ccaa68e2e110782a580f1553f549b074e112222139037f3ce5d24641a8f5993"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2604.23354","created_at":"2026-05-29T01:05:10.514818+00:00"},{"alias_kind":"arxiv_version","alias_value":"2604.23354v2","created_at":"2026-05-29T01:05:10.514818+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.23354","created_at":"2026-05-29T01:05:10.514818+00:00"},{"alias_kind":"pith_short_12","alias_value":"7ECUX2WT3FVS","created_at":"2026-05-29T01:05:10.514818+00:00"},{"alias_kind":"pith_short_16","alias_value":"7ECUX2WT3FVS2IHK","created_at":"2026-05-29T01:05:10.514818+00:00"},{"alias_kind":"pith_short_8","alias_value":"7ECUX2WT","created_at":"2026-05-29T01:05:10.514818+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7ECUX2WT3FVS2IHK4OZBKP6AL6","json":"https://pith.science/pith/7ECUX2WT3FVS2IHK4OZBKP6AL6.json","graph_json":"https://pith.science/api/pith-number/7ECUX2WT3FVS2IHK4OZBKP6AL6/graph.json","events_json":"https://pith.science/api/pith-number/7ECUX2WT3FVS2IHK4OZBKP6AL6/events.json","paper":"https://pith.science/paper/7ECUX2WT"},"agent_actions":{"view_html":"https://pith.science/pith/7ECUX2WT3FVS2IHK4OZBKP6AL6","download_json":"https://pith.science/pith/7ECUX2WT3FVS2IHK4OZBKP6AL6.json","view_paper":"https://pith.science/paper/7ECUX2WT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2604.23354&json=true","fetch_graph":"https://pith.science/api/pith-number/7ECUX2WT3FVS2IHK4OZBKP6AL6/graph.json","fetch_events":"https://pith.science/api/pith-number/7ECUX2WT3FVS2IHK4OZBKP6AL6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7ECUX2WT3FVS2IHK4OZBKP6AL6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7ECUX2WT3FVS2IHK4OZBKP6AL6/action/storage_attestation","attest_author":"https://pith.science/pith/7ECUX2WT3FVS2IHK4OZBKP6AL6/action/author_attestation","sign_citation":"https://pith.science/pith/7ECUX2WT3FVS2IHK4OZBKP6AL6/action/citation_signature","submit_replication":"https://pith.science/pith/7ECUX2WT3FVS2IHK4OZBKP6AL6/action/replication_record"}},"created_at":"2026-05-29T01:05:10.514818+00:00","updated_at":"2026-05-29T01:05:10.514818+00:00"}