{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:2AWDZ7VJ7DTO4UXSOOSWDYA5AP","short_pith_number":"pith:2AWDZ7VJ","schema_version":"1.0","canonical_sha256":"d02c3cfea9f8e6ee52f273a561e01d03caed465613e32fa0c84ad4b96c1c77d8","source":{"kind":"arxiv","id":"2310.00672","version":2},"attestation_state":"computed","paper":{"title":"GeRA: Label-Efficient Geometrically Regularized Alignment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.CV"],"primary_cat":"cs.LG","authors_text":"Dustin Klebe, Justin Solomon, Leonid Karlinsky, Mikhail Yurochkin, Tal Shnitzer","submitted_at":"2023-10-01T13:48:36Z","abstract_excerpt":"Pretrained unimodal encoders incorporate rich semantic information into embedding space structures. To be similarly informative, multi-modal encoders typically require massive amounts of paired data for alignment and training. We introduce a semi-supervised Geometrically Regularized Alignment (GeRA) method to align the embedding spaces of pretrained unimodal encoders in a label-efficient way. Our method leverages the manifold geometry of unpaired (unlabeled) data to improve alignment performance. To prevent distortions to local geometry during the alignment process, potentially disrupting sema"},"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":"2310.00672","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-10-01T13:48:36Z","cross_cats_sorted":["cs.CL","cs.CV"],"title_canon_sha256":"7458d47ec718c256b5d79ff804a84a83ae5c99f001f2c93b554c4cf38c16b5e1","abstract_canon_sha256":"58eb7e0d0069092a1eb7cb8f2ccd8e4f3ba3b48c4e773eecc38bbfcc1196142c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:58:17.711583Z","signature_b64":"QUFJc4WHru5lBMZAXB6k2JQAIn/stAwc6vbV8E8WK6CjKDwE1zJEUqmDVLVkxwgwEZQLiDVR/bK2UazxnT6TBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d02c3cfea9f8e6ee52f273a561e01d03caed465613e32fa0c84ad4b96c1c77d8","last_reissued_at":"2026-07-05T06:58:17.711066Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:58:17.711066Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"GeRA: Label-Efficient Geometrically Regularized Alignment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.CV"],"primary_cat":"cs.LG","authors_text":"Dustin Klebe, Justin Solomon, Leonid Karlinsky, Mikhail Yurochkin, Tal Shnitzer","submitted_at":"2023-10-01T13:48:36Z","abstract_excerpt":"Pretrained unimodal encoders incorporate rich semantic information into embedding space structures. To be similarly informative, multi-modal encoders typically require massive amounts of paired data for alignment and training. We introduce a semi-supervised Geometrically Regularized Alignment (GeRA) method to align the embedding spaces of pretrained unimodal encoders in a label-efficient way. Our method leverages the manifold geometry of unpaired (unlabeled) data to improve alignment performance. To prevent distortions to local geometry during the alignment process, potentially disrupting sema"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.00672","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/2310.00672/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":"2310.00672","created_at":"2026-07-05T06:58:17.711136+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.00672v2","created_at":"2026-07-05T06:58:17.711136+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.00672","created_at":"2026-07-05T06:58:17.711136+00:00"},{"alias_kind":"pith_short_12","alias_value":"2AWDZ7VJ7DTO","created_at":"2026-07-05T06:58:17.711136+00:00"},{"alias_kind":"pith_short_16","alias_value":"2AWDZ7VJ7DTO4UXS","created_at":"2026-07-05T06:58:17.711136+00:00"},{"alias_kind":"pith_short_8","alias_value":"2AWDZ7VJ","created_at":"2026-07-05T06:58:17.711136+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.12322","citing_title":"Model alignment using inter-modal bridges","ref_index":2014,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2AWDZ7VJ7DTO4UXSOOSWDYA5AP","json":"https://pith.science/pith/2AWDZ7VJ7DTO4UXSOOSWDYA5AP.json","graph_json":"https://pith.science/api/pith-number/2AWDZ7VJ7DTO4UXSOOSWDYA5AP/graph.json","events_json":"https://pith.science/api/pith-number/2AWDZ7VJ7DTO4UXSOOSWDYA5AP/events.json","paper":"https://pith.science/paper/2AWDZ7VJ"},"agent_actions":{"view_html":"https://pith.science/pith/2AWDZ7VJ7DTO4UXSOOSWDYA5AP","download_json":"https://pith.science/pith/2AWDZ7VJ7DTO4UXSOOSWDYA5AP.json","view_paper":"https://pith.science/paper/2AWDZ7VJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.00672&json=true","fetch_graph":"https://pith.science/api/pith-number/2AWDZ7VJ7DTO4UXSOOSWDYA5AP/graph.json","fetch_events":"https://pith.science/api/pith-number/2AWDZ7VJ7DTO4UXSOOSWDYA5AP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2AWDZ7VJ7DTO4UXSOOSWDYA5AP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2AWDZ7VJ7DTO4UXSOOSWDYA5AP/action/storage_attestation","attest_author":"https://pith.science/pith/2AWDZ7VJ7DTO4UXSOOSWDYA5AP/action/author_attestation","sign_citation":"https://pith.science/pith/2AWDZ7VJ7DTO4UXSOOSWDYA5AP/action/citation_signature","submit_replication":"https://pith.science/pith/2AWDZ7VJ7DTO4UXSOOSWDYA5AP/action/replication_record"}},"created_at":"2026-07-05T06:58:17.711136+00:00","updated_at":"2026-07-05T06:58:17.711136+00:00"}