{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:TNPJH5PUXL7LUZV5AWFDVDQQYK","short_pith_number":"pith:TNPJH5PU","schema_version":"1.0","canonical_sha256":"9b5e93f5f4bafeba66bd058a3a8e10c29d757d334cfad4fcb7d12e4a2aca7aa8","source":{"kind":"arxiv","id":"2410.02035","version":1},"attestation_state":"computed","paper":{"title":"Tuning Frequency Bias of State Space Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Annan Yu, Dongwei Lyu, Michael W. Mahoney, N. Benjamin Erichson, Soon Hoe Lim","submitted_at":"2024-10-02T21:04:22Z","abstract_excerpt":"State space models (SSMs) leverage linear, time-invariant (LTI) systems to effectively learn sequences with long-range dependencies. By analyzing the transfer functions of LTI systems, we find that SSMs exhibit an implicit bias toward capturing low-frequency components more effectively than high-frequency ones. This behavior aligns with the broader notion of frequency bias in deep learning model training. We show that the initialization of an SSM assigns it an innate frequency bias and that training the model in a conventional way does not alter this bias. Based on our theory, we propose two m"},"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":"2410.02035","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-10-02T21:04:22Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"d9896962952564346746ac3557602f7e873a18ea9d58394adac0e529093b3387","abstract_canon_sha256":"15bb17fbfef4b19e580010752c9d4ab8fea5741f0b83fc5269c228379da86300"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:14:59.452104Z","signature_b64":"xq3NELnMNNgtb9t1ZXkblA1K6mia5Llbi2ZKJ64jk3oJIo+94RpN6YELEk/dDH6ljQSR4xmoudkphme8se4JBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9b5e93f5f4bafeba66bd058a3a8e10c29d757d334cfad4fcb7d12e4a2aca7aa8","last_reissued_at":"2026-07-05T09:14:59.451680Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:14:59.451680Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Tuning Frequency Bias of State Space Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Annan Yu, Dongwei Lyu, Michael W. Mahoney, N. Benjamin Erichson, Soon Hoe Lim","submitted_at":"2024-10-02T21:04:22Z","abstract_excerpt":"State space models (SSMs) leverage linear, time-invariant (LTI) systems to effectively learn sequences with long-range dependencies. By analyzing the transfer functions of LTI systems, we find that SSMs exhibit an implicit bias toward capturing low-frequency components more effectively than high-frequency ones. This behavior aligns with the broader notion of frequency bias in deep learning model training. We show that the initialization of an SSM assigns it an innate frequency bias and that training the model in a conventional way does not alter this bias. Based on our theory, we propose two m"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.02035","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/2410.02035/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":"2410.02035","created_at":"2026-07-05T09:14:59.451737+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.02035v1","created_at":"2026-07-05T09:14:59.451737+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.02035","created_at":"2026-07-05T09:14:59.451737+00:00"},{"alias_kind":"pith_short_12","alias_value":"TNPJH5PUXL7L","created_at":"2026-07-05T09:14:59.451737+00:00"},{"alias_kind":"pith_short_16","alias_value":"TNPJH5PUXL7LUZV5","created_at":"2026-07-05T09:14:59.451737+00:00"},{"alias_kind":"pith_short_8","alias_value":"TNPJH5PU","created_at":"2026-07-05T09:14:59.451737+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.11089","citing_title":"Structured State-Space Regularization for Generation-Friendly Image Tokenization","ref_index":65,"is_internal_anchor":false},{"citing_arxiv_id":"2602.08318","citing_title":"Is Flow Matching Just Trajectory Replay for Sequential Data?","ref_index":105,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08539","citing_title":"Continuity Laws for Sequential Models","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2604.11089","citing_title":"Structured State-Space Regularization for Generation-Friendly Image Tokenization","ref_index":65,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TNPJH5PUXL7LUZV5AWFDVDQQYK","json":"https://pith.science/pith/TNPJH5PUXL7LUZV5AWFDVDQQYK.json","graph_json":"https://pith.science/api/pith-number/TNPJH5PUXL7LUZV5AWFDVDQQYK/graph.json","events_json":"https://pith.science/api/pith-number/TNPJH5PUXL7LUZV5AWFDVDQQYK/events.json","paper":"https://pith.science/paper/TNPJH5PU"},"agent_actions":{"view_html":"https://pith.science/pith/TNPJH5PUXL7LUZV5AWFDVDQQYK","download_json":"https://pith.science/pith/TNPJH5PUXL7LUZV5AWFDVDQQYK.json","view_paper":"https://pith.science/paper/TNPJH5PU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.02035&json=true","fetch_graph":"https://pith.science/api/pith-number/TNPJH5PUXL7LUZV5AWFDVDQQYK/graph.json","fetch_events":"https://pith.science/api/pith-number/TNPJH5PUXL7LUZV5AWFDVDQQYK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TNPJH5PUXL7LUZV5AWFDVDQQYK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TNPJH5PUXL7LUZV5AWFDVDQQYK/action/storage_attestation","attest_author":"https://pith.science/pith/TNPJH5PUXL7LUZV5AWFDVDQQYK/action/author_attestation","sign_citation":"https://pith.science/pith/TNPJH5PUXL7LUZV5AWFDVDQQYK/action/citation_signature","submit_replication":"https://pith.science/pith/TNPJH5PUXL7LUZV5AWFDVDQQYK/action/replication_record"}},"created_at":"2026-07-05T09:14:59.451737+00:00","updated_at":"2026-07-05T09:14:59.451737+00:00"}