{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:Z3MN3YHFXKHP3KRB4ZZVHW4JA6","short_pith_number":"pith:Z3MN3YHF","schema_version":"1.0","canonical_sha256":"ced8dde0e5ba8efdaa21e67353db8907813696aaf3f8b6be7e51ec7aaeafdc0b","source":{"kind":"arxiv","id":"2605.11490","version":2},"attestation_state":"computed","paper":{"title":"Adaptive Calibration in Non-Stationary Environments","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Online algorithms achieve calibration errors that adapt to the unknown degree of non-stationarity in the outcome sequence.","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Haipeng Luo, Junyan Liu, Lillian J. Ratliff","submitted_at":"2026-05-12T04:06:40Z","abstract_excerpt":"Making calibrated online predictions is a central challenge in modern AI systems. Much of the existing literature focuses on fully adversarial environments where outcomes may be arbitrary, leading to conservative algorithms that can perform suboptimally in more benign settings, such as when outcomes are nearly stationary. This gap raises a natural question: can we design online prediction algorithms whose calibration error automatically adapts to the degree of non-stationarity in the environment, smoothly interpolating between i.i.d. and adversarial regimes? We answer this question in the affi"},"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":"2605.11490","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-12T04:06:40Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"652bb2e580b68c23aa1ee55a250cbd1422f8e052e4e993dc37ef35be28587f49","abstract_canon_sha256":"f77f7aa49fa0be81992d67f2e15b1ff58d58f26a49d88c3eb5d7b2fa3c8cfdde"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-25T02:01:23.164438Z","signature_b64":"SpQ5Ow7RStQ2OVzk2Mo/KeP21W0qp+jLzgnU6mGaJo2tXL0MsV+Y6jmobTrisI3a3Guh49kF1oQsjmxB0Sj4CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ced8dde0e5ba8efdaa21e67353db8907813696aaf3f8b6be7e51ec7aaeafdc0b","last_reissued_at":"2026-05-25T02:01:23.163616Z","signature_status":"signed_v1","first_computed_at":"2026-05-25T02:01:23.163616Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Adaptive Calibration in Non-Stationary Environments","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Online algorithms achieve calibration errors that adapt to the unknown degree of non-stationarity in the outcome sequence.","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Haipeng Luo, Junyan Liu, Lillian J. Ratliff","submitted_at":"2026-05-12T04:06:40Z","abstract_excerpt":"Making calibrated online predictions is a central challenge in modern AI systems. Much of the existing literature focuses on fully adversarial environments where outcomes may be arbitrary, leading to conservative algorithms that can perform suboptimally in more benign settings, such as when outcomes are nearly stationary. This gap raises a natural question: can we design online prediction algorithms whose calibration error automatically adapts to the degree of non-stationarity in the environment, smoothly interpolating between i.i.d. and adversarial regimes? We answer this question in the affi"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"our algorithms attain Õ(√T+(TC)^{1/3}) for ℓ1 calibration error and Õ((1+C)^{1/3}) for both ℓ2 and pseudo KL calibration error. These bounds match the optimal rates in the stationary case (C=0) and recover known guarantees in the fully adversarial regime (C=T).","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The non-stationarity of the environment can be captured by the minimal ℓ1 deviation C of mean outcomes, and epoch-based scheduling combined with a non-uniform partition of the prediction space suffices to achieve the stated adaptive bounds without knowledge of C.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Online algorithms achieve adaptive calibration bounds of Õ(√T + (TC)^{1/3}) for ℓ1 error and Õ((1+C)^{1/3}) for ℓ2 and pseudo-KL error, matching stationary and adversarial extremes via epoch scheduling and non-uniform prediction partitioning.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Online algorithms achieve calibration errors that adapt to the unknown degree of non-stationarity in the outcome sequence.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"e4832a9d63f3c482a03ad7690866241e2e63bc2b96ae194c15a8b7a90778a40d"},"source":{"id":"2605.11490","kind":"arxiv","version":2},"verdict":{"id":"748e6ecc-508c-4e91-aae5-e5fa5dce7138","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-13T01:43:02.653670Z","strongest_claim":"our algorithms attain Õ(√T+(TC)^{1/3}) for ℓ1 calibration error and Õ((1+C)^{1/3}) for both ℓ2 and pseudo KL calibration error. These bounds match the optimal rates in the stationary case (C=0) and recover known guarantees in the fully adversarial regime (C=T).","one_line_summary":"Online algorithms achieve adaptive calibration bounds of Õ(√T + (TC)^{1/3}) for ℓ1 error and Õ((1+C)^{1/3}) for ℓ2 and pseudo-KL error, matching stationary and adversarial extremes via epoch scheduling and non-uniform prediction partitioning.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The non-stationarity of the environment can be captured by the minimal ℓ1 deviation C of mean outcomes, and epoch-based scheduling combined with a non-uniform partition of the prediction space suffices to achieve the stated adaptive bounds without knowledge of C.","pith_extraction_headline":"Online algorithms achieve calibration errors that adapt to the unknown degree of non-stationarity in the outcome sequence."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.11490/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"claim_evidence","ran_at":"2026-05-20T04:02:00.733312Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T12:35:01.778851Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-19T09:31:19.231732Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T08:21:04.852024Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"78c4f355738c4c0934b0fefa04fb6223f54f5e86e8400dfa8666c62fa8ddf3c8"},"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":"2605.11490","created_at":"2026-05-25T02:01:23.163746+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.11490v2","created_at":"2026-05-25T02:01:23.163746+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.11490","created_at":"2026-05-25T02:01:23.163746+00:00"},{"alias_kind":"pith_short_12","alias_value":"Z3MN3YHFXKHP","created_at":"2026-05-25T02:01:23.163746+00:00"},{"alias_kind":"pith_short_16","alias_value":"Z3MN3YHFXKHP3KRB","created_at":"2026-05-25T02:01:23.163746+00:00"},{"alias_kind":"pith_short_8","alias_value":"Z3MN3YHF","created_at":"2026-05-25T02:01:23.163746+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.09273","citing_title":"Instance-Adaptive Online Multicalibration","ref_index":110,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/Z3MN3YHFXKHP3KRB4ZZVHW4JA6","json":"https://pith.science/pith/Z3MN3YHFXKHP3KRB4ZZVHW4JA6.json","graph_json":"https://pith.science/api/pith-number/Z3MN3YHFXKHP3KRB4ZZVHW4JA6/graph.json","events_json":"https://pith.science/api/pith-number/Z3MN3YHFXKHP3KRB4ZZVHW4JA6/events.json","paper":"https://pith.science/paper/Z3MN3YHF"},"agent_actions":{"view_html":"https://pith.science/pith/Z3MN3YHFXKHP3KRB4ZZVHW4JA6","download_json":"https://pith.science/pith/Z3MN3YHFXKHP3KRB4ZZVHW4JA6.json","view_paper":"https://pith.science/paper/Z3MN3YHF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.11490&json=true","fetch_graph":"https://pith.science/api/pith-number/Z3MN3YHFXKHP3KRB4ZZVHW4JA6/graph.json","fetch_events":"https://pith.science/api/pith-number/Z3MN3YHFXKHP3KRB4ZZVHW4JA6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Z3MN3YHFXKHP3KRB4ZZVHW4JA6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Z3MN3YHFXKHP3KRB4ZZVHW4JA6/action/storage_attestation","attest_author":"https://pith.science/pith/Z3MN3YHFXKHP3KRB4ZZVHW4JA6/action/author_attestation","sign_citation":"https://pith.science/pith/Z3MN3YHFXKHP3KRB4ZZVHW4JA6/action/citation_signature","submit_replication":"https://pith.science/pith/Z3MN3YHFXKHP3KRB4ZZVHW4JA6/action/replication_record"}},"created_at":"2026-05-25T02:01:23.163746+00:00","updated_at":"2026-05-25T02:01:23.163746+00:00"}