{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:OAQGB7OG2LMJ7PHCJEQQVRPFSV","short_pith_number":"pith:OAQGB7OG","schema_version":"1.0","canonical_sha256":"702060fdc6d2d89fbce249210ac5e595543f2ee74351d3e108a27391f1455a00","source":{"kind":"arxiv","id":"2606.12691","version":1},"attestation_state":"computed","paper":{"title":"Two-Layer Linear Auto-Regressive Models Estimate Latent States","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.SY","eess.SY","math.OC","stat.ML"],"primary_cat":"cs.LG","authors_text":"Leo Maynard-Zhang, Maryam Fazel, Sarah Dean, Sunmook Choi, Yahya Sattar, Yassir Jedra","submitted_at":"2026-06-10T21:25:16Z","abstract_excerpt":"Auto-regressive models have emerged as powerful tools for sequential data, from language to video. Understanding how and why these models learn latent representations remains an open theoretical question. In this work, we demonstrate that when trained by empirical risk minimization on data from partially observed linear dynamical systems, two-layer linear auto-regressive models naturally learn to approximate Kalman filtering. In particular, we show that the learned hidden representation coincides, up to a similarity transformation, with the state estimates produced by the optimal (Kalman) filt"},"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":"2606.12691","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-10T21:25:16Z","cross_cats_sorted":["cs.AI","cs.SY","eess.SY","math.OC","stat.ML"],"title_canon_sha256":"e7d2d1353967eb9d5379dbb78bdcbe9cd8de9f0d257f80e4142cc7bd4823c304","abstract_canon_sha256":"4d668f8cfb4d7b1735f3aa7024ca12e0b2de057c9f083902e8c15ba4b11377a9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-12T01:08:45.696007Z","signature_b64":"X+RQkstP8xTDfEP4Mp9RaioIHI8Nl9xKH7nndocbbIBYiqfUsS0CEBZz4tLjEiNvPTmTbO7B5hZ8HgOjjx/rCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"702060fdc6d2d89fbce249210ac5e595543f2ee74351d3e108a27391f1455a00","last_reissued_at":"2026-06-12T01:08:45.695037Z","signature_status":"signed_v1","first_computed_at":"2026-06-12T01:08:45.695037Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Two-Layer Linear Auto-Regressive Models Estimate Latent States","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.SY","eess.SY","math.OC","stat.ML"],"primary_cat":"cs.LG","authors_text":"Leo Maynard-Zhang, Maryam Fazel, Sarah Dean, Sunmook Choi, Yahya Sattar, Yassir Jedra","submitted_at":"2026-06-10T21:25:16Z","abstract_excerpt":"Auto-regressive models have emerged as powerful tools for sequential data, from language to video. Understanding how and why these models learn latent representations remains an open theoretical question. In this work, we demonstrate that when trained by empirical risk minimization on data from partially observed linear dynamical systems, two-layer linear auto-regressive models naturally learn to approximate Kalman filtering. In particular, we show that the learned hidden representation coincides, up to a similarity transformation, with the state estimates produced by the optimal (Kalman) filt"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.12691","kind":"arxiv","version":1},"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/2606.12691/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":"2606.12691","created_at":"2026-06-12T01:08:45.695196+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.12691v1","created_at":"2026-06-12T01:08:45.695196+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.12691","created_at":"2026-06-12T01:08:45.695196+00:00"},{"alias_kind":"pith_short_12","alias_value":"OAQGB7OG2LMJ","created_at":"2026-06-12T01:08:45.695196+00:00"},{"alias_kind":"pith_short_16","alias_value":"OAQGB7OG2LMJ7PHC","created_at":"2026-06-12T01:08:45.695196+00:00"},{"alias_kind":"pith_short_8","alias_value":"OAQGB7OG","created_at":"2026-06-12T01:08:45.695196+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OAQGB7OG2LMJ7PHCJEQQVRPFSV","json":"https://pith.science/pith/OAQGB7OG2LMJ7PHCJEQQVRPFSV.json","graph_json":"https://pith.science/api/pith-number/OAQGB7OG2LMJ7PHCJEQQVRPFSV/graph.json","events_json":"https://pith.science/api/pith-number/OAQGB7OG2LMJ7PHCJEQQVRPFSV/events.json","paper":"https://pith.science/paper/OAQGB7OG"},"agent_actions":{"view_html":"https://pith.science/pith/OAQGB7OG2LMJ7PHCJEQQVRPFSV","download_json":"https://pith.science/pith/OAQGB7OG2LMJ7PHCJEQQVRPFSV.json","view_paper":"https://pith.science/paper/OAQGB7OG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.12691&json=true","fetch_graph":"https://pith.science/api/pith-number/OAQGB7OG2LMJ7PHCJEQQVRPFSV/graph.json","fetch_events":"https://pith.science/api/pith-number/OAQGB7OG2LMJ7PHCJEQQVRPFSV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OAQGB7OG2LMJ7PHCJEQQVRPFSV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OAQGB7OG2LMJ7PHCJEQQVRPFSV/action/storage_attestation","attest_author":"https://pith.science/pith/OAQGB7OG2LMJ7PHCJEQQVRPFSV/action/author_attestation","sign_citation":"https://pith.science/pith/OAQGB7OG2LMJ7PHCJEQQVRPFSV/action/citation_signature","submit_replication":"https://pith.science/pith/OAQGB7OG2LMJ7PHCJEQQVRPFSV/action/replication_record"}},"created_at":"2026-06-12T01:08:45.695196+00:00","updated_at":"2026-06-12T01:08:45.695196+00:00"}