{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:435PGUIDUC6DZSMGBCPYNJQ66N","short_pith_number":"pith:435PGUID","schema_version":"1.0","canonical_sha256":"e6faf35103a0bc3cc986089f86a61ef3511797a45d1dea86f6918ce7e156196c","source":{"kind":"arxiv","id":"2312.07930","version":3},"attestation_state":"computed","paper":{"title":"Towards Optimal Statistical Watermarking","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.CR","cs.IT","math.IT","stat.ML"],"primary_cat":"cs.LG","authors_text":"Baihe Huang, Banghua Zhu, Hanlin Zhu, Jason D. Lee, Jiantao Jiao, Kannan Ramchandran, Michael I. Jordan","submitted_at":"2023-12-13T06:57:00Z","abstract_excerpt":"We study statistical watermarking by formulating it as a hypothesis testing problem, a general framework which subsumes all previous statistical watermarking methods. Key to our formulation is a coupling of the output tokens and the rejection region, realized by pseudo-random generators in practice, that allows non-trivial trade-offs between the Type I error and Type II error. We characterize the Uniformly Most Powerful (UMP) watermark in the general hypothesis testing setting and the minimax Type II error in the model-agnostic setting. In the common scenario where the output is a sequence of "},"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":"2312.07930","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-12-13T06:57:00Z","cross_cats_sorted":["cs.CL","cs.CR","cs.IT","math.IT","stat.ML"],"title_canon_sha256":"2350d0f00b90bbda4633da9d19572f53729f406a1af325380d7e6b495f83a9df","abstract_canon_sha256":"51440defb6585ccb27f7c3477d21524c89276c768e55da49f4375f7b93512bcf"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:42:22.247242Z","signature_b64":"DaMvAd3HrTDMYBA9cm1dcP+0+GKlVoaDk/uUFK8hjsrdtOMSGnWOmT+LMcomPGXLftyo7ZEugo9rrYV2DbGDBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e6faf35103a0bc3cc986089f86a61ef3511797a45d1dea86f6918ce7e156196c","last_reissued_at":"2026-07-05T07:42:22.246722Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:42:22.246722Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Towards Optimal Statistical Watermarking","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.CR","cs.IT","math.IT","stat.ML"],"primary_cat":"cs.LG","authors_text":"Baihe Huang, Banghua Zhu, Hanlin Zhu, Jason D. Lee, Jiantao Jiao, Kannan Ramchandran, Michael I. Jordan","submitted_at":"2023-12-13T06:57:00Z","abstract_excerpt":"We study statistical watermarking by formulating it as a hypothesis testing problem, a general framework which subsumes all previous statistical watermarking methods. Key to our formulation is a coupling of the output tokens and the rejection region, realized by pseudo-random generators in practice, that allows non-trivial trade-offs between the Type I error and Type II error. We characterize the Uniformly Most Powerful (UMP) watermark in the general hypothesis testing setting and the minimax Type II error in the model-agnostic setting. In the common scenario where the output is a sequence of "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2312.07930","kind":"arxiv","version":3},"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/2312.07930/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":"2312.07930","created_at":"2026-07-05T07:42:22.246789+00:00"},{"alias_kind":"arxiv_version","alias_value":"2312.07930v3","created_at":"2026-07-05T07:42:22.246789+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2312.07930","created_at":"2026-07-05T07:42:22.246789+00:00"},{"alias_kind":"pith_short_12","alias_value":"435PGUIDUC6D","created_at":"2026-07-05T07:42:22.246789+00:00"},{"alias_kind":"pith_short_16","alias_value":"435PGUIDUC6DZSMG","created_at":"2026-07-05T07:42:22.246789+00:00"},{"alias_kind":"pith_short_8","alias_value":"435PGUID","created_at":"2026-07-05T07:42:22.246789+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.10977","citing_title":"PASA: A Principled Embedding-Space Watermarking Approach for LLM-Generated Text under Semantic-Invariant Attacks","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10977","citing_title":"PASA: A Principled Embedding-Space Watermarking Approach for LLM-Generated Text under Semantic-Invariant Attacks","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08826","citing_title":"Fundamental Trade-Offs in Multi-Bit Watermarking of Stochastic Processes","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08964","citing_title":"Trustworthy AI: Ensuring Reliability and Accountability from Models to Agents","ref_index":79,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/435PGUIDUC6DZSMGBCPYNJQ66N","json":"https://pith.science/pith/435PGUIDUC6DZSMGBCPYNJQ66N.json","graph_json":"https://pith.science/api/pith-number/435PGUIDUC6DZSMGBCPYNJQ66N/graph.json","events_json":"https://pith.science/api/pith-number/435PGUIDUC6DZSMGBCPYNJQ66N/events.json","paper":"https://pith.science/paper/435PGUID"},"agent_actions":{"view_html":"https://pith.science/pith/435PGUIDUC6DZSMGBCPYNJQ66N","download_json":"https://pith.science/pith/435PGUIDUC6DZSMGBCPYNJQ66N.json","view_paper":"https://pith.science/paper/435PGUID","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2312.07930&json=true","fetch_graph":"https://pith.science/api/pith-number/435PGUIDUC6DZSMGBCPYNJQ66N/graph.json","fetch_events":"https://pith.science/api/pith-number/435PGUIDUC6DZSMGBCPYNJQ66N/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/435PGUIDUC6DZSMGBCPYNJQ66N/action/timestamp_anchor","attest_storage":"https://pith.science/pith/435PGUIDUC6DZSMGBCPYNJQ66N/action/storage_attestation","attest_author":"https://pith.science/pith/435PGUIDUC6DZSMGBCPYNJQ66N/action/author_attestation","sign_citation":"https://pith.science/pith/435PGUIDUC6DZSMGBCPYNJQ66N/action/citation_signature","submit_replication":"https://pith.science/pith/435PGUIDUC6DZSMGBCPYNJQ66N/action/replication_record"}},"created_at":"2026-07-05T07:42:22.246789+00:00","updated_at":"2026-07-05T07:42:22.246789+00:00"}