{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:D3RWSQW746Q35EB2BHJW4QGQ43","short_pith_number":"pith:D3RWSQW7","schema_version":"1.0","canonical_sha256":"1ee36942dfe7a1be903a09d36e40d0e6f043099d4addb596b00b562fd881166e","source":{"kind":"arxiv","id":"2603.19910","version":2},"attestation_state":"computed","paper":{"title":"Learning Adaptive Parameter Policies for Nonlinear Bayesian Filtering","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Reinforcement learning trains policies to choose filter parameters dynamically in nonlinear Bayesian estimation.","cross_cats":["cs.SY"],"primary_cat":"eess.SY","authors_text":"Felipe Giraldo-Grueso, Ondrej Straka, Renato Zanetti","submitted_at":"2026-03-20T12:48:26Z","abstract_excerpt":"For many nonlinear Bayesian state estimation problems, the posterior recursion is not analytically tractable, leading to algorithms that are influenced by numerical approximation errors. These algorithms depend on parameters that affect the approximation's accuracy and computational cost. The parameters include, for example, the number of particles, scaling parameters, and the number of iterations in iterative computations. Typically, these parameters are fixed or adjusted heuristically, although the approximation accuracy can change over time with the local degree of nonlinearity and uncertai"},"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":"2603.19910","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"eess.SY","submitted_at":"2026-03-20T12:48:26Z","cross_cats_sorted":["cs.SY"],"title_canon_sha256":"73a19d27ce8cdbdf98a4c8691807c9827fcbfff0ee5e6c584c2cb7d0068a5edf","abstract_canon_sha256":"80a3c93c5d47e4ae76ce5194ac663a7c187e840b6ab196e2d742fa2a6106634d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T03:09:22.643819Z","signature_b64":"tYBAG7FlVS71pd0SYIOdzQbD3wSB/V/fUVG74Cy7lSegIFkCdGGq+CPk5Z4Yr7FUfFt+Flm+SlWv5mhBFNLzAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1ee36942dfe7a1be903a09d36e40d0e6f043099d4addb596b00b562fd881166e","last_reissued_at":"2026-05-18T03:09:22.643051Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T03:09:22.643051Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning Adaptive Parameter Policies for Nonlinear Bayesian Filtering","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Reinforcement learning trains policies to choose filter parameters dynamically in nonlinear Bayesian estimation.","cross_cats":["cs.SY"],"primary_cat":"eess.SY","authors_text":"Felipe Giraldo-Grueso, Ondrej Straka, Renato Zanetti","submitted_at":"2026-03-20T12:48:26Z","abstract_excerpt":"For many nonlinear Bayesian state estimation problems, the posterior recursion is not analytically tractable, leading to algorithms that are influenced by numerical approximation errors. These algorithms depend on parameters that affect the approximation's accuracy and computational cost. The parameters include, for example, the number of particles, scaling parameters, and the number of iterations in iterative computations. Typically, these parameters are fixed or adjusted heuristically, although the approximation accuracy can change over time with the local degree of nonlinearity and uncertai"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Experiments with the unscented Kalman filter and stochastic integration filter demonstrate that the learned policies improve both estimate quality and consistency.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That a reward function defined on estimation accuracy and consistency will produce policies that generalize beyond the training scenarios to real-world time-varying nonlinearity without retraining or instability.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Reinforcement learning is used to learn adaptive policies for selecting parameters in nonlinear Bayesian filters, improving estimate quality and consistency in experiments with the unscented Kalman filter and stochastic integration filter.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Reinforcement learning trains policies to choose filter parameters dynamically in nonlinear Bayesian estimation.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"7894ae160b9bbe4ae666fcf9fa221caf4643c4bb5373660387aae025be53a4a7"},"source":{"id":"2603.19910","kind":"arxiv","version":2},"verdict":{"id":"c2de5849-8ad7-4e1a-80d8-4f868167716d","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T08:39:18.924732Z","strongest_claim":"Experiments with the unscented Kalman filter and stochastic integration filter demonstrate that the learned policies improve both estimate quality and consistency.","one_line_summary":"Reinforcement learning is used to learn adaptive policies for selecting parameters in nonlinear Bayesian filters, improving estimate quality and consistency in experiments with the unscented Kalman filter and stochastic integration filter.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That a reward function defined on estimation accuracy and consistency will produce policies that generalize beyond the training scenarios to real-world time-varying nonlinearity without retraining or instability.","pith_extraction_headline":"Reinforcement learning trains policies to choose filter parameters dynamically in nonlinear Bayesian estimation."},"references":{"count":33,"sample":[{"doi":"","year":2013,"title":"S ¨arkk¨a,Bayesian Filtering and Smoothing","work_id":"7423d4c6-4e32-4eb5-bca0-bf2c7fba141e","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2004,"title":"Unscented filtering and nonlinear estimation,","work_id":"a2d582e7-aed4-49fd-8ea4-610ecb215cac","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2013,"title":"J. Dun ´ık, O. Straka, and M.ˇSimandl, “Stochastic integration filter,”IEEE Transactions on Automatic Control, vol. 58, no. 6, pp. 1561–1566, 2013","work_id":"909e67bf-2049-4aa2-bf87-03d93d5269d0","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1994,"title":"Sequential data assimilation with a nonlinear quasi- geostrophic model using Monte Carlo methods to forecast error statis- tics,","work_id":"60572691-f2cb-424f-b9f5-1d46aa51e72c","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1999,"title":"A Monte Carlo implementation of the nonlinear filtering problem to produce ensemble assimilations and forecasts,","work_id":"e47393af-eb8e-4901-b5d6-ca2ed6a8bd76","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":33,"snapshot_sha256":"9bbabff84fe9aed98f1ab5240f59e2ac1d965cb841d98ce48a08c2176197eeb6","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"caf7cb41932e041a58aaba208634994eba3626eb83e8ecd3891ea6c112b914e1"},"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":"2603.19910","created_at":"2026-05-18T03:09:22.643184+00:00"},{"alias_kind":"arxiv_version","alias_value":"2603.19910v2","created_at":"2026-05-18T03:09:22.643184+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.19910","created_at":"2026-05-18T03:09:22.643184+00:00"},{"alias_kind":"pith_short_12","alias_value":"D3RWSQW746Q3","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"D3RWSQW746Q35EB2","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"D3RWSQW7","created_at":"2026-05-18T12:33:37.589309+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/D3RWSQW746Q35EB2BHJW4QGQ43","json":"https://pith.science/pith/D3RWSQW746Q35EB2BHJW4QGQ43.json","graph_json":"https://pith.science/api/pith-number/D3RWSQW746Q35EB2BHJW4QGQ43/graph.json","events_json":"https://pith.science/api/pith-number/D3RWSQW746Q35EB2BHJW4QGQ43/events.json","paper":"https://pith.science/paper/D3RWSQW7"},"agent_actions":{"view_html":"https://pith.science/pith/D3RWSQW746Q35EB2BHJW4QGQ43","download_json":"https://pith.science/pith/D3RWSQW746Q35EB2BHJW4QGQ43.json","view_paper":"https://pith.science/paper/D3RWSQW7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2603.19910&json=true","fetch_graph":"https://pith.science/api/pith-number/D3RWSQW746Q35EB2BHJW4QGQ43/graph.json","fetch_events":"https://pith.science/api/pith-number/D3RWSQW746Q35EB2BHJW4QGQ43/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/D3RWSQW746Q35EB2BHJW4QGQ43/action/timestamp_anchor","attest_storage":"https://pith.science/pith/D3RWSQW746Q35EB2BHJW4QGQ43/action/storage_attestation","attest_author":"https://pith.science/pith/D3RWSQW746Q35EB2BHJW4QGQ43/action/author_attestation","sign_citation":"https://pith.science/pith/D3RWSQW746Q35EB2BHJW4QGQ43/action/citation_signature","submit_replication":"https://pith.science/pith/D3RWSQW746Q35EB2BHJW4QGQ43/action/replication_record"}},"created_at":"2026-05-18T03:09:22.643184+00:00","updated_at":"2026-05-18T03:09:22.643184+00:00"}