{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:CCXCMYOFNDKGYJULI3E2MAZ4QL","short_pith_number":"pith:CCXCMYOF","schema_version":"1.0","canonical_sha256":"10ae2661c568d46c268b46c9a6033c82e5632bb3a8aa66c7af5d364eda4e913d","source":{"kind":"arxiv","id":"2410.06303","version":3},"attestation_state":"computed","paper":{"title":"Compositional Risk Minimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Charles Arnal, Divyat Mahajan, Ioannis Mitliagkas, Kartik Ahuja, Mohammad Pezeshki, Pascal Vincent","submitted_at":"2024-10-08T19:25:07Z","abstract_excerpt":"Compositional generalization is a crucial step towards developing data-efficient intelligent machines that generalize in human-like ways. In this work, we tackle a challenging form of distribution shift, termed compositional shift, where some attribute combinations are completely absent at training but present in the test distribution. This shift tests the model's ability to generalize compositionally to novel attribute combinations in discriminative tasks. We model the data with flexible additive energy distributions, where each energy term represents an attribute, and derive a simple alterna"},"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":"2410.06303","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-10-08T19:25:07Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"17b3ad787f7c1c6bca6b3fd62dd10a2afd3a6505b2786ebd930adf37b7d757f9","abstract_canon_sha256":"fe9cbf62983b36409858320a844c6708e094869d4477146235781148d867783e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:35:10.804444Z","signature_b64":"5VixPWT3GQwImzcuGT4l0O2Z4mBWKFwz/c4yysxSfAOTZeIbpPGAJalJvlLRC70NWeoCjkmAdPBcUMx2QSzbBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"10ae2661c568d46c268b46c9a6033c82e5632bb3a8aa66c7af5d364eda4e913d","last_reissued_at":"2026-07-05T11:35:10.803958Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:35:10.803958Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Compositional Risk Minimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Charles Arnal, Divyat Mahajan, Ioannis Mitliagkas, Kartik Ahuja, Mohammad Pezeshki, Pascal Vincent","submitted_at":"2024-10-08T19:25:07Z","abstract_excerpt":"Compositional generalization is a crucial step towards developing data-efficient intelligent machines that generalize in human-like ways. In this work, we tackle a challenging form of distribution shift, termed compositional shift, where some attribute combinations are completely absent at training but present in the test distribution. This shift tests the model's ability to generalize compositionally to novel attribute combinations in discriminative tasks. We model the data with flexible additive energy distributions, where each energy term represents an attribute, and derive a simple alterna"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.06303","kind":"arxiv","version":3},"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/2410.06303/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":"2410.06303","created_at":"2026-07-05T11:35:10.804012+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.06303v3","created_at":"2026-07-05T11:35:10.804012+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.06303","created_at":"2026-07-05T11:35:10.804012+00:00"},{"alias_kind":"pith_short_12","alias_value":"CCXCMYOFNDKG","created_at":"2026-07-05T11:35:10.804012+00:00"},{"alias_kind":"pith_short_16","alias_value":"CCXCMYOFNDKGYJUL","created_at":"2026-07-05T11:35:10.804012+00:00"},{"alias_kind":"pith_short_8","alias_value":"CCXCMYOF","created_at":"2026-07-05T11:35:10.804012+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.24161","citing_title":"Dual-Branch Cross-Projection Debiasing through Diffusion-based Disentanglement","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2606.21646","citing_title":"Energy-based Compositional Diffusion Planning","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2606.19594","citing_title":"Unsupervised Causal Abstractions Discovery","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11170","citing_title":"Unlearning with Asymmetric Sources: Improved Unlearning-Utility Trade-off with Public Data","ref_index":133,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18258","citing_title":"Long-Text-to-Image Generation via Compositional Prompt Decomposition","ref_index":19,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CCXCMYOFNDKGYJULI3E2MAZ4QL","json":"https://pith.science/pith/CCXCMYOFNDKGYJULI3E2MAZ4QL.json","graph_json":"https://pith.science/api/pith-number/CCXCMYOFNDKGYJULI3E2MAZ4QL/graph.json","events_json":"https://pith.science/api/pith-number/CCXCMYOFNDKGYJULI3E2MAZ4QL/events.json","paper":"https://pith.science/paper/CCXCMYOF"},"agent_actions":{"view_html":"https://pith.science/pith/CCXCMYOFNDKGYJULI3E2MAZ4QL","download_json":"https://pith.science/pith/CCXCMYOFNDKGYJULI3E2MAZ4QL.json","view_paper":"https://pith.science/paper/CCXCMYOF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.06303&json=true","fetch_graph":"https://pith.science/api/pith-number/CCXCMYOFNDKGYJULI3E2MAZ4QL/graph.json","fetch_events":"https://pith.science/api/pith-number/CCXCMYOFNDKGYJULI3E2MAZ4QL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CCXCMYOFNDKGYJULI3E2MAZ4QL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CCXCMYOFNDKGYJULI3E2MAZ4QL/action/storage_attestation","attest_author":"https://pith.science/pith/CCXCMYOFNDKGYJULI3E2MAZ4QL/action/author_attestation","sign_citation":"https://pith.science/pith/CCXCMYOFNDKGYJULI3E2MAZ4QL/action/citation_signature","submit_replication":"https://pith.science/pith/CCXCMYOFNDKGYJULI3E2MAZ4QL/action/replication_record"}},"created_at":"2026-07-05T11:35:10.804012+00:00","updated_at":"2026-07-05T11:35:10.804012+00:00"}