{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:JH45OEVIQGZ3U2DVFPQRQ4KIAK","short_pith_number":"pith:JH45OEVI","schema_version":"1.0","canonical_sha256":"49f9d712a881b3ba68752be118714802a3c096c949f3a77cc361c749f51babe2","source":{"kind":"arxiv","id":"2605.16818","version":1},"attestation_state":"computed","paper":{"title":"Observation-Aligned Mask Priors for Learning Physical Dynamics from Authentic Occlusions","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Mask priors learned from authentic occlusions create context-query splits that give every observed dimension a positive chance of being queried.","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Chiyuan Ma, Tianshu Yu, Zihan Zhou","submitted_at":"2026-05-16T05:23:49Z","abstract_excerpt":"Learning physical dynamics directly from incomplete observations is challenging because authentic occlusions are structured, sample-dependent, and often missing not at random, whereas existing methods typically rely on heuristic masking rules or predefined mask distributions. We propose Observation-Aligned Mask Priors, a framework that learns the distribution of authentic observation masks and uses it to construct context-query partitions for training from incomplete data. Specifically, we pretrain a Bayesian Flow Network (BFN) on binary observation masks to capture real occlusion topologies, "},"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":true,"formal_links_present":true},"canonical_record":{"source":{"id":"2605.16818","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-16T05:23:49Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"4efddd94f0aaa4740b97f24d8c12b42a170afce4d1691fb1313b902a82ef7d24","abstract_canon_sha256":"614c860d5a539587de24d4569e4ea2a118b9f1efe4462058df222a8213cac853"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:03:24.110822Z","signature_b64":"EZPMvrpsezPJ/HOirORkSQ7anXM7QFcm6koLamycJ5VB0SKFtOfKyhvbZ7vIii2q6vEzd6/zx9M7Ndrr/hNxDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"49f9d712a881b3ba68752be118714802a3c096c949f3a77cc361c749f51babe2","last_reissued_at":"2026-05-20T00:03:24.110010Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:03:24.110010Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Observation-Aligned Mask Priors for Learning Physical Dynamics from Authentic Occlusions","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Mask priors learned from authentic occlusions create context-query splits that give every observed dimension a positive chance of being queried.","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Chiyuan Ma, Tianshu Yu, Zihan Zhou","submitted_at":"2026-05-16T05:23:49Z","abstract_excerpt":"Learning physical dynamics directly from incomplete observations is challenging because authentic occlusions are structured, sample-dependent, and often missing not at random, whereas existing methods typically rely on heuristic masking rules or predefined mask distributions. We propose Observation-Aligned Mask Priors, a framework that learns the distribution of authentic observation masks and uses it to construct context-query partitions for training from incomplete data. Specifically, we pretrain a Bayesian Flow Network (BFN) on binary observation masks to capture real occlusion topologies, "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"This intersection-based partitioning gives every valid observed dimension a strictly positive probability of being queried, preventing zero-query dead zones and local generative collapse.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The pretrained Bayesian Flow Network on binary observation masks from the target datasets accurately captures the true distribution of authentic occlusions and that the globally normalized cross-entropy guidance produces sample-specific masks without introducing systematic bias in the resulting context-query splits.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A framework pretrained on authentic binary occlusion masks uses guided sampling and intersection-based partitioning to train diffusion models on incomplete physical observations without zero-query regions.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Mask priors learned from authentic occlusions create context-query splits that give every observed dimension a positive chance of being queried.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"c0739ffe71ada06428b4ba0464c806d53130eb7a367258e8d1278ba93946a195"},"source":{"id":"2605.16818","kind":"arxiv","version":1},"verdict":{"id":"fd7b050b-b822-42ad-bc91-f03bbaa4f8af","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T20:55:17.201730Z","strongest_claim":"This intersection-based partitioning gives every valid observed dimension a strictly positive probability of being queried, preventing zero-query dead zones and local generative collapse.","one_line_summary":"A framework pretrained on authentic binary occlusion masks uses guided sampling and intersection-based partitioning to train diffusion models on incomplete physical observations without zero-query regions.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The pretrained Bayesian Flow Network on binary observation masks from the target datasets accurately captures the true distribution of authentic occlusions and that the globally normalized cross-entropy guidance produces sample-specific masks without introducing systematic bias in the resulting context-query splits.","pith_extraction_headline":"Mask priors learned from authentic occlusions create context-query splits that give every observed dimension a positive chance of being queried."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.16818/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_compliance","ran_at":"2026-05-19T21:01:26.876852Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-19T21:01:19.267746Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T19:01:56.271336Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T18:33:26.411661Z","status":"skipped","version":"1.0.0","findings_count":0}],"snapshot_sha256":"a0999fbc28feb8c3d785f661f7eacc392c405133d923687d4c1e012e6a91665a"},"references":{"count":36,"sample":[{"doi":"","year":2024,"title":"Alexander Barth, Julien Brajard, Aida Alvera-Azcárate, Bayoumy Mohamed, Charles Troupin, and Jean-Marie Beckers. Ensemble reconstruction of missing satellite data using a denoising diffusion model: ap","work_id":"cb932b15-062e-415d-9101-c04d747b4fb8","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Generative data assimilation for surface ocean state estimation from multi-modal satellite observations.Journal of Advances in Modeling Earth Systems, 17(8):e2025MS005063, 2025","work_id":"77204fc5-2f1f-4b8a-aef1-a1235a8d088f","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Accurate medium-range global weather forecasting with 3d neural networks.Nature, 619(7970):533–538","work_id":"c09c22c3-989d-4467-897d-9b6dc0d2eae9","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Learning skillful medium-range global weather forecasting.Science, 382(6677):1416–1421","work_id":"947ca872-4d7a-47d3-a939-2008d7f25295","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Forecasting corporate financial performance using deep learning with environmental, social, and governance data","work_id":"09885817-3bc7-446b-9f12-76a76d84d208","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":36,"snapshot_sha256":"3ad9d9475ec6da2b3b09cbd1ab77fd4e176157944f6c28c1b872c049246da213","internal_anchors":3},"formal_canon":{"evidence_count":2,"snapshot_sha256":"196df5ddb718ebd87a4dfe495b2836586fc3388a9ce4a5cfd1d207aa1260c125"},"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":"2605.16818","created_at":"2026-05-20T00:03:24.110142+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.16818v1","created_at":"2026-05-20T00:03:24.110142+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.16818","created_at":"2026-05-20T00:03:24.110142+00:00"},{"alias_kind":"pith_short_12","alias_value":"JH45OEVIQGZ3","created_at":"2026-05-20T00:03:24.110142+00:00"},{"alias_kind":"pith_short_16","alias_value":"JH45OEVIQGZ3U2DV","created_at":"2026-05-20T00:03:24.110142+00:00"},{"alias_kind":"pith_short_8","alias_value":"JH45OEVI","created_at":"2026-05-20T00:03:24.110142+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JH45OEVIQGZ3U2DVFPQRQ4KIAK","json":"https://pith.science/pith/JH45OEVIQGZ3U2DVFPQRQ4KIAK.json","graph_json":"https://pith.science/api/pith-number/JH45OEVIQGZ3U2DVFPQRQ4KIAK/graph.json","events_json":"https://pith.science/api/pith-number/JH45OEVIQGZ3U2DVFPQRQ4KIAK/events.json","paper":"https://pith.science/paper/JH45OEVI"},"agent_actions":{"view_html":"https://pith.science/pith/JH45OEVIQGZ3U2DVFPQRQ4KIAK","download_json":"https://pith.science/pith/JH45OEVIQGZ3U2DVFPQRQ4KIAK.json","view_paper":"https://pith.science/paper/JH45OEVI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.16818&json=true","fetch_graph":"https://pith.science/api/pith-number/JH45OEVIQGZ3U2DVFPQRQ4KIAK/graph.json","fetch_events":"https://pith.science/api/pith-number/JH45OEVIQGZ3U2DVFPQRQ4KIAK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JH45OEVIQGZ3U2DVFPQRQ4KIAK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JH45OEVIQGZ3U2DVFPQRQ4KIAK/action/storage_attestation","attest_author":"https://pith.science/pith/JH45OEVIQGZ3U2DVFPQRQ4KIAK/action/author_attestation","sign_citation":"https://pith.science/pith/JH45OEVIQGZ3U2DVFPQRQ4KIAK/action/citation_signature","submit_replication":"https://pith.science/pith/JH45OEVIQGZ3U2DVFPQRQ4KIAK/action/replication_record"}},"created_at":"2026-05-20T00:03:24.110142+00:00","updated_at":"2026-05-20T00:03:24.110142+00:00"}