{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:Z2PUSD437NOAFNYG4BLWBN3547","short_pith_number":"pith:Z2PUSD43","schema_version":"1.0","canonical_sha256":"ce9f490f9bfb5c02b706e05760b77de7c8135f8f0a783b9a9c22806714fcba84","source":{"kind":"arxiv","id":"2605.26012","version":1},"attestation_state":"computed","paper":{"title":"Learning in Low-Dimensional Subspaces: Orthogonal Bottlenecks for Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Aleksandar Todorov, Matthia Sabatelli","submitted_at":"2026-05-25T16:31:33Z","abstract_excerpt":"Deep reinforcement learning (RL) agents commonly rely on high-dimensional neural representations, despite growing evidence that task-relevant value and policy structure may be intrinsically low-dimensional. In this work, we present a simple yet effective representation-level prior that inserts a fixed orthonormal projection to constrain encoder features to a low-dimensional subspace, requiring no auxiliary objectives, pretraining, or changes to the underlying RL algorithm. Under a linear realizability assumption, we prove that when the bottleneck dimension exceeds the intrinsic rank of the opt"},"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":"2605.26012","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-25T16:31:33Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"e9230357fbdda8d0a79ab241afd724b48ea20b2b9ff79f4ef7723a0b140e29f1","abstract_canon_sha256":"c41629f7b2d1c7e946f8fe1ba466f094cb1604b1ba0dfdbb0d7eaecc10792447"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T02:05:23.216540Z","signature_b64":"WxFlXa9X1j7FiRaOFfapRA5cGUbXRCIfBonUdwOS2gc/rn589+BW46QkMbqjTssthL8TUzVLGUk8QKFaoHHQAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ce9f490f9bfb5c02b706e05760b77de7c8135f8f0a783b9a9c22806714fcba84","last_reissued_at":"2026-05-26T02:05:23.215935Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T02:05:23.215935Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning in Low-Dimensional Subspaces: Orthogonal Bottlenecks for Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Aleksandar Todorov, Matthia Sabatelli","submitted_at":"2026-05-25T16:31:33Z","abstract_excerpt":"Deep reinforcement learning (RL) agents commonly rely on high-dimensional neural representations, despite growing evidence that task-relevant value and policy structure may be intrinsically low-dimensional. In this work, we present a simple yet effective representation-level prior that inserts a fixed orthonormal projection to constrain encoder features to a low-dimensional subspace, requiring no auxiliary objectives, pretraining, or changes to the underlying RL algorithm. Under a linear realizability assumption, we prove that when the bottleneck dimension exceeds the intrinsic rank of the opt"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.26012","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/2605.26012/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":"2605.26012","created_at":"2026-05-26T02:05:23.216016+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.26012v1","created_at":"2026-05-26T02:05:23.216016+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.26012","created_at":"2026-05-26T02:05:23.216016+00:00"},{"alias_kind":"pith_short_12","alias_value":"Z2PUSD437NOA","created_at":"2026-05-26T02:05:23.216016+00:00"},{"alias_kind":"pith_short_16","alias_value":"Z2PUSD437NOAFNYG","created_at":"2026-05-26T02:05:23.216016+00:00"},{"alias_kind":"pith_short_8","alias_value":"Z2PUSD43","created_at":"2026-05-26T02:05:23.216016+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/Z2PUSD437NOAFNYG4BLWBN3547","json":"https://pith.science/pith/Z2PUSD437NOAFNYG4BLWBN3547.json","graph_json":"https://pith.science/api/pith-number/Z2PUSD437NOAFNYG4BLWBN3547/graph.json","events_json":"https://pith.science/api/pith-number/Z2PUSD437NOAFNYG4BLWBN3547/events.json","paper":"https://pith.science/paper/Z2PUSD43"},"agent_actions":{"view_html":"https://pith.science/pith/Z2PUSD437NOAFNYG4BLWBN3547","download_json":"https://pith.science/pith/Z2PUSD437NOAFNYG4BLWBN3547.json","view_paper":"https://pith.science/paper/Z2PUSD43","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.26012&json=true","fetch_graph":"https://pith.science/api/pith-number/Z2PUSD437NOAFNYG4BLWBN3547/graph.json","fetch_events":"https://pith.science/api/pith-number/Z2PUSD437NOAFNYG4BLWBN3547/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Z2PUSD437NOAFNYG4BLWBN3547/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Z2PUSD437NOAFNYG4BLWBN3547/action/storage_attestation","attest_author":"https://pith.science/pith/Z2PUSD437NOAFNYG4BLWBN3547/action/author_attestation","sign_citation":"https://pith.science/pith/Z2PUSD437NOAFNYG4BLWBN3547/action/citation_signature","submit_replication":"https://pith.science/pith/Z2PUSD437NOAFNYG4BLWBN3547/action/replication_record"}},"created_at":"2026-05-26T02:05:23.216016+00:00","updated_at":"2026-05-26T02:05:23.216016+00:00"}