{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:H2RCZW3UZP75XOLDKATW4NB7TI","short_pith_number":"pith:H2RCZW3U","schema_version":"1.0","canonical_sha256":"3ea22cdb74cbffdbb96350276e343f9a22fc142ca6e873bab7ca71f3691984a4","source":{"kind":"arxiv","id":"2103.14608","version":2},"attestation_state":"computed","paper":{"title":"On UMAP's true loss function","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Fred A. Hamprecht, Sebastian Damrich","submitted_at":"2021-03-26T17:22:58Z","abstract_excerpt":"UMAP has supplanted t-SNE as state-of-the-art for visualizing high-dimensional datasets in many disciplines, but the reason for its success is not well understood. In this work, we investigate UMAP's sampling based optimization scheme in detail. We derive UMAP's effective loss function in closed form and find that it differs from the published one. As a consequence, we show that UMAP does not aim to reproduce its theoretically motivated high-dimensional UMAP similarities. Instead, it tries to reproduce similarities that only encode the shared $k$ nearest neighbor graph, thereby challenging the"},"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":"2103.14608","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-03-26T17:22:58Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"89fa3f7a952f6c12e1f86f19bbeef388d8732d07b8c8e52c2c29b4f581049eb2","abstract_canon_sha256":"ff20e38d4d6358edb409506eb2e3af42baeadc3c0a8e27e673d7d746eb054928"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:34:12.246291Z","signature_b64":"XXFS4Fx6szneku2qFeJTEo4xyAVuolfcsycM6SfJHgQQ/0cwDWavTfYRj/oKrClR/N80bH5cN7f50XmilqH9Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3ea22cdb74cbffdbb96350276e343f9a22fc142ca6e873bab7ca71f3691984a4","last_reissued_at":"2026-07-05T02:34:12.245670Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:34:12.245670Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"On UMAP's true loss function","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Fred A. Hamprecht, Sebastian Damrich","submitted_at":"2021-03-26T17:22:58Z","abstract_excerpt":"UMAP has supplanted t-SNE as state-of-the-art for visualizing high-dimensional datasets in many disciplines, but the reason for its success is not well understood. In this work, we investigate UMAP's sampling based optimization scheme in detail. We derive UMAP's effective loss function in closed form and find that it differs from the published one. As a consequence, we show that UMAP does not aim to reproduce its theoretically motivated high-dimensional UMAP similarities. Instead, it tries to reproduce similarities that only encode the shared $k$ nearest neighbor graph, thereby challenging the"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2103.14608","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2103.14608/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":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":"2103.14608","created_at":"2026-07-05T02:34:12.245749+00:00"},{"alias_kind":"arxiv_version","alias_value":"2103.14608v2","created_at":"2026-07-05T02:34:12.245749+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2103.14608","created_at":"2026-07-05T02:34:12.245749+00:00"},{"alias_kind":"pith_short_12","alias_value":"H2RCZW3UZP75","created_at":"2026-07-05T02:34:12.245749+00:00"},{"alias_kind":"pith_short_16","alias_value":"H2RCZW3UZP75XOLD","created_at":"2026-07-05T02:34:12.245749+00:00"},{"alias_kind":"pith_short_8","alias_value":"H2RCZW3U","created_at":"2026-07-05T02:34:12.245749+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2608.11269","citing_title":"CosMAP: Contrastive Manifold Approximation and Projection for Dimensionality Reduction of Omics and Genealogical Data","ref_index":79,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/H2RCZW3UZP75XOLDKATW4NB7TI","json":"https://pith.science/pith/H2RCZW3UZP75XOLDKATW4NB7TI.json","graph_json":"https://pith.science/api/pith-number/H2RCZW3UZP75XOLDKATW4NB7TI/graph.json","events_json":"https://pith.science/api/pith-number/H2RCZW3UZP75XOLDKATW4NB7TI/events.json","paper":"https://pith.science/paper/H2RCZW3U"},"agent_actions":{"view_html":"https://pith.science/pith/H2RCZW3UZP75XOLDKATW4NB7TI","download_json":"https://pith.science/pith/H2RCZW3UZP75XOLDKATW4NB7TI.json","view_paper":"https://pith.science/paper/H2RCZW3U","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2103.14608&json=true","fetch_graph":"https://pith.science/api/pith-number/H2RCZW3UZP75XOLDKATW4NB7TI/graph.json","fetch_events":"https://pith.science/api/pith-number/H2RCZW3UZP75XOLDKATW4NB7TI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/H2RCZW3UZP75XOLDKATW4NB7TI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/H2RCZW3UZP75XOLDKATW4NB7TI/action/storage_attestation","attest_author":"https://pith.science/pith/H2RCZW3UZP75XOLDKATW4NB7TI/action/author_attestation","sign_citation":"https://pith.science/pith/H2RCZW3UZP75XOLDKATW4NB7TI/action/citation_signature","submit_replication":"https://pith.science/pith/H2RCZW3UZP75XOLDKATW4NB7TI/action/replication_record"}},"created_at":"2026-07-05T02:34:12.245749+00:00","updated_at":"2026-07-05T02:34:12.245749+00:00"}