{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:EOBLTWW5FKJKYZKHDV5ALXTBQG","short_pith_number":"pith:EOBLTWW5","schema_version":"1.0","canonical_sha256":"2382b9dadd2a92ac65471d7a05de61818ff03ac13db519682a36c0a89c1414ef","source":{"kind":"arxiv","id":"2509.14431","version":2},"attestation_state":"computed","paper":{"title":"Local-Canonicalization Equivariant Graph Neural Networks for Sample-Efficient and Generalizable Swarm Robot Control","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.RO","authors_text":"Christine Allen-Blanchette, David Chang, Keqin Wang, Tao Zhong","submitted_at":"2025-09-17T21:11:05Z","abstract_excerpt":"Multi-agent reinforcement learning (MARL) policies for swarm control often learn inefficiently and generalize poorly across coordinate frames, team sizes, and agent roles. We introduce Local-Canonicalization Equivariant Graph Neural Networks (LEGO), a modular policy architecture that combines agent-centric canonicalization with role-aware graph encoding. Canonicalization removes dependence on the global coordinate frame, while transforming predicted local actions back to the world frame produces an E(2)-equivariant policy. Role-wise graph encoders provide intra-role permutation equivariance an"},"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":"2509.14431","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.RO","submitted_at":"2025-09-17T21:11:05Z","cross_cats_sorted":[],"title_canon_sha256":"53ec7d3a83207ab9fc68bb2988d7b947ae5383816337260b759a0e73cf3c2a98","abstract_canon_sha256":"8287555a82fa61c2929654bc7c057df0f42b33bf6df8330eb1893ff20140c7b2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-04T00:30:17.732477Z","signature_b64":"u60k3/r7+9TCNO1RylKAYszxyLlzcZjI0YMCbJ8Mvd0B0QZP0kK0SUdZsgN6JuRPk41kh/ujU3E/lX8GwbNuAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2382b9dadd2a92ac65471d7a05de61818ff03ac13db519682a36c0a89c1414ef","last_reissued_at":"2026-08-04T00:30:17.730727Z","signature_status":"signed_v1","first_computed_at":"2026-08-04T00:30:17.730727Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Local-Canonicalization Equivariant Graph Neural Networks for Sample-Efficient and Generalizable Swarm Robot Control","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.RO","authors_text":"Christine Allen-Blanchette, David Chang, Keqin Wang, Tao Zhong","submitted_at":"2025-09-17T21:11:05Z","abstract_excerpt":"Multi-agent reinforcement learning (MARL) policies for swarm control often learn inefficiently and generalize poorly across coordinate frames, team sizes, and agent roles. We introduce Local-Canonicalization Equivariant Graph Neural Networks (LEGO), a modular policy architecture that combines agent-centric canonicalization with role-aware graph encoding. Canonicalization removes dependence on the global coordinate frame, while transforming predicted local actions back to the world frame produces an E(2)-equivariant policy. Role-wise graph encoders provide intra-role permutation equivariance an"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.14431","kind":"arxiv","version":2},"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/2509.14431/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":"2509.14431","created_at":"2026-08-04T00:30:17.731782+00:00"},{"alias_kind":"arxiv_version","alias_value":"2509.14431v2","created_at":"2026-08-04T00:30:17.731782+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.14431","created_at":"2026-08-04T00:30:17.731782+00:00"},{"alias_kind":"pith_short_12","alias_value":"EOBLTWW5FKJK","created_at":"2026-08-04T00:30:17.731782+00:00"},{"alias_kind":"pith_short_16","alias_value":"EOBLTWW5FKJKYZKH","created_at":"2026-08-04T00:30:17.731782+00:00"},{"alias_kind":"pith_short_8","alias_value":"EOBLTWW5","created_at":"2026-08-04T00:30:17.731782+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.25867","citing_title":"CINOC: Cardinality-Invariant Neural Operator Policies for Scalable PDE Control","ref_index":5,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/EOBLTWW5FKJKYZKHDV5ALXTBQG","json":"https://pith.science/pith/EOBLTWW5FKJKYZKHDV5ALXTBQG.json","graph_json":"https://pith.science/api/pith-number/EOBLTWW5FKJKYZKHDV5ALXTBQG/graph.json","events_json":"https://pith.science/api/pith-number/EOBLTWW5FKJKYZKHDV5ALXTBQG/events.json","paper":"https://pith.science/paper/EOBLTWW5"},"agent_actions":{"view_html":"https://pith.science/pith/EOBLTWW5FKJKYZKHDV5ALXTBQG","download_json":"https://pith.science/pith/EOBLTWW5FKJKYZKHDV5ALXTBQG.json","view_paper":"https://pith.science/paper/EOBLTWW5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2509.14431&json=true","fetch_graph":"https://pith.science/api/pith-number/EOBLTWW5FKJKYZKHDV5ALXTBQG/graph.json","fetch_events":"https://pith.science/api/pith-number/EOBLTWW5FKJKYZKHDV5ALXTBQG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/EOBLTWW5FKJKYZKHDV5ALXTBQG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/EOBLTWW5FKJKYZKHDV5ALXTBQG/action/storage_attestation","attest_author":"https://pith.science/pith/EOBLTWW5FKJKYZKHDV5ALXTBQG/action/author_attestation","sign_citation":"https://pith.science/pith/EOBLTWW5FKJKYZKHDV5ALXTBQG/action/citation_signature","submit_replication":"https://pith.science/pith/EOBLTWW5FKJKYZKHDV5ALXTBQG/action/replication_record"}},"created_at":"2026-08-04T00:30:17.731782+00:00","updated_at":"2026-08-04T00:30:17.731782+00:00"}