{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:JGPL3UEPS7ZFCI5QLILN5I54N4","short_pith_number":"pith:JGPL3UEP","schema_version":"1.0","canonical_sha256":"499ebdd08f97f25123b05a16dea3bc6f287bed26bc18bbfc315c44ef80f6307d","source":{"kind":"arxiv","id":"2603.08558","version":3},"attestation_state":"computed","paper":{"title":"Impact of Connectivity on Laplacian Representations in Reinforcement Learning","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"Linear value approximation error using learned Laplacian spectral features is upper-bounded by the algebraic connectivity of the MDP state graph.","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Keyue Jiang, Laura Toni, Matteo Papini, Pierriccardo Olivieri, Tommaso Giorgi","submitted_at":"2026-03-09T16:20:31Z","abstract_excerpt":"Learning compact state representations in Markov Decision Processes (MDPs) has proven crucial for addressing the curse of dimensionality in large-scale reinforcement learning (RL) problems. Existing principled approaches leverage structural priors on the MDP by constructing state representations as linear combinations of the state-graph Laplacian eigenvectors. When the transition graph is unknown or the state space is prohibitively large, the graph spectral features can be estimated directly via sample trajectories. In this work, we prove an upper bound on the approximation error of linear val"},"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":true},"canonical_record":{"source":{"id":"2603.08558","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2026-03-09T16:20:31Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"91b69987c6553d1f8627588aca8f30d2ee7999f9fd8c02b02fd995a8e8927a17","abstract_canon_sha256":"680ff75cdf20e2980cef475fdb034404aab0028d0a88a1b5dd6dcf719ab58bd5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-11T01:10:34.739102Z","signature_b64":"BsSRHNDfMq3QK0cTqsGuZgk9HBk3uxvduFYjutvtr/sAO2/RyXdnwnXwjEd5Ha7+soeiOXyQnyUldjxlwBTNBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"499ebdd08f97f25123b05a16dea3bc6f287bed26bc18bbfc315c44ef80f6307d","last_reissued_at":"2026-06-11T01:10:34.738030Z","signature_status":"signed_v1","first_computed_at":"2026-06-11T01:10:34.738030Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Impact of Connectivity on Laplacian Representations in Reinforcement Learning","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"Linear value approximation error using learned Laplacian spectral features is upper-bounded by the algebraic connectivity of the MDP state graph.","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Keyue Jiang, Laura Toni, Matteo Papini, Pierriccardo Olivieri, Tommaso Giorgi","submitted_at":"2026-03-09T16:20:31Z","abstract_excerpt":"Learning compact state representations in Markov Decision Processes (MDPs) has proven crucial for addressing the curse of dimensionality in large-scale reinforcement learning (RL) problems. Existing principled approaches leverage structural priors on the MDP by constructing state representations as linear combinations of the state-graph Laplacian eigenvectors. When the transition graph is unknown or the state space is prohibitively large, the graph spectral features can be estimated directly via sample trajectories. In this work, we prove an upper bound on the approximation error of linear val"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We prove an upper bound on the approximation error of linear value function approximation under the learned spectral features. We show how this error scales with the algebraic connectivity of the state-graph, grounding the approximation quality in the topological structure of the MDP.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The transition graph is treated as undirected or symmetrizable for the Laplacian construction, even though the paper states results hold without symmetry assumptions on the induced kernel; the algebraic connectivity measure may implicitly require a symmetric or reversible structure for the standard Laplacian definition used.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"The approximation error for linear value functions using learned Laplacian spectral features is upper-bounded by the algebraic connectivity of the MDP transition graph, with an additional bound on eigenvector estimation error from trajectories.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Linear value approximation error using learned Laplacian spectral features is upper-bounded by the algebraic connectivity of the MDP state graph.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"b480d4ade2535a5061c9f9bab3ef12388e42d7067a1c2ba9f000732089ce4962"},"source":{"id":"2603.08558","kind":"arxiv","version":3},"verdict":{"id":"16534f17-74e1-46db-bbcf-e27abeaa0d2d","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T14:36:48.191722Z","strongest_claim":"We prove an upper bound on the approximation error of linear value function approximation under the learned spectral features. We show how this error scales with the algebraic connectivity of the state-graph, grounding the approximation quality in the topological structure of the MDP.","one_line_summary":"The approximation error for linear value functions using learned Laplacian spectral features is upper-bounded by the algebraic connectivity of the MDP transition graph, with an additional bound on eigenvector estimation error from trajectories.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The transition graph is treated as undirected or symmetrizable for the Laplacian construction, even though the paper states results hold without symmetry assumptions on the induced kernel; the algebraic connectivity measure may implicitly require a symmetric or reversible structure for the standard Laplacian definition used.","pith_extraction_headline":"Linear value approximation error using learned Laplacian spectral features is upper-bounded by the algebraic connectivity of the MDP state graph."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2603.08558/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":2,"snapshot_sha256":"cecb84ca8ca04b78f58aea9b253399b5f9e5500910268aa946749df4481c6a50"},"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":"2603.08558","created_at":"2026-06-11T01:10:34.738159+00:00"},{"alias_kind":"arxiv_version","alias_value":"2603.08558v3","created_at":"2026-06-11T01:10:34.738159+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.08558","created_at":"2026-06-11T01:10:34.738159+00:00"},{"alias_kind":"pith_short_12","alias_value":"JGPL3UEPS7ZF","created_at":"2026-06-11T01:10:34.738159+00:00"},{"alias_kind":"pith_short_16","alias_value":"JGPL3UEPS7ZFCI5Q","created_at":"2026-06-11T01:10:34.738159+00:00"},{"alias_kind":"pith_short_8","alias_value":"JGPL3UEP","created_at":"2026-06-11T01:10:34.738159+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/JGPL3UEPS7ZFCI5QLILN5I54N4","json":"https://pith.science/pith/JGPL3UEPS7ZFCI5QLILN5I54N4.json","graph_json":"https://pith.science/api/pith-number/JGPL3UEPS7ZFCI5QLILN5I54N4/graph.json","events_json":"https://pith.science/api/pith-number/JGPL3UEPS7ZFCI5QLILN5I54N4/events.json","paper":"https://pith.science/paper/JGPL3UEP"},"agent_actions":{"view_html":"https://pith.science/pith/JGPL3UEPS7ZFCI5QLILN5I54N4","download_json":"https://pith.science/pith/JGPL3UEPS7ZFCI5QLILN5I54N4.json","view_paper":"https://pith.science/paper/JGPL3UEP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2603.08558&json=true","fetch_graph":"https://pith.science/api/pith-number/JGPL3UEPS7ZFCI5QLILN5I54N4/graph.json","fetch_events":"https://pith.science/api/pith-number/JGPL3UEPS7ZFCI5QLILN5I54N4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JGPL3UEPS7ZFCI5QLILN5I54N4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JGPL3UEPS7ZFCI5QLILN5I54N4/action/storage_attestation","attest_author":"https://pith.science/pith/JGPL3UEPS7ZFCI5QLILN5I54N4/action/author_attestation","sign_citation":"https://pith.science/pith/JGPL3UEPS7ZFCI5QLILN5I54N4/action/citation_signature","submit_replication":"https://pith.science/pith/JGPL3UEPS7ZFCI5QLILN5I54N4/action/replication_record"}},"created_at":"2026-06-11T01:10:34.738159+00:00","updated_at":"2026-06-11T01:10:34.738159+00:00"}