{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:CR3FIJISUIVV6A6MDPY4C634R5","short_pith_number":"pith:CR3FIJIS","schema_version":"1.0","canonical_sha256":"1476542512a22b5f03cc1bf1c17b7c8f77d1b6fa93e754652a0655b40b2f3cde","source":{"kind":"arxiv","id":"2004.05007","version":2},"attestation_state":"computed","paper":{"title":"An Empirical Study of Invariant Risk Minimization","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Jiyeon Ham, Kyubyong Park, Yo Joong Choe","submitted_at":"2020-04-10T12:23:29Z","abstract_excerpt":"Invariant risk minimization (IRM) (Arjovsky et al., 2019) is a recently proposed framework designed for learning predictors that are invariant to spurious correlations across different training environments. Yet, despite its theoretical justifications, IRM has not been extensively tested across various settings. In an attempt to gain a better understanding of the framework, we empirically investigate several research questions using IRMv1, which is the first practical algorithm proposed to approximately solve IRM. By extending the ColoredMNIST experiment in different ways, we find that IRMv1 ("},"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":"2004.05007","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"stat.ML","submitted_at":"2020-04-10T12:23:29Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"f6d9305bbfb45c699b2c1afb401a9bee7ebbdd8332e06f2d1dcdcca199ec7125","abstract_canon_sha256":"c592aba3d61f52320d96c4e884c6cb3547ed8c44c916374329be1774da951ca3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:16:14.484991Z","signature_b64":"PnCAog9s3SAPFiD0A5NB9ap/JsHVZK2GgkJyPEvqIWXmF52IVNsizKuZ65rYQhpI8hCtBVqyX/1FQHYr0vK9CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1476542512a22b5f03cc1bf1c17b7c8f77d1b6fa93e754652a0655b40b2f3cde","last_reissued_at":"2026-07-05T01:16:14.484417Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:16:14.484417Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"An Empirical Study of Invariant Risk Minimization","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Jiyeon Ham, Kyubyong Park, Yo Joong Choe","submitted_at":"2020-04-10T12:23:29Z","abstract_excerpt":"Invariant risk minimization (IRM) (Arjovsky et al., 2019) is a recently proposed framework designed for learning predictors that are invariant to spurious correlations across different training environments. Yet, despite its theoretical justifications, IRM has not been extensively tested across various settings. In an attempt to gain a better understanding of the framework, we empirically investigate several research questions using IRMv1, which is the first practical algorithm proposed to approximately solve IRM. By extending the ColoredMNIST experiment in different ways, we find that IRMv1 ("},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2004.05007","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/2004.05007/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":"2004.05007","created_at":"2026-07-05T01:16:14.484496+00:00"},{"alias_kind":"arxiv_version","alias_value":"2004.05007v2","created_at":"2026-07-05T01:16:14.484496+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2004.05007","created_at":"2026-07-05T01:16:14.484496+00:00"},{"alias_kind":"pith_short_12","alias_value":"CR3FIJISUIVV","created_at":"2026-07-05T01:16:14.484496+00:00"},{"alias_kind":"pith_short_16","alias_value":"CR3FIJISUIVV6A6M","created_at":"2026-07-05T01:16:14.484496+00:00"},{"alias_kind":"pith_short_8","alias_value":"CR3FIJIS","created_at":"2026-07-05T01:16:14.484496+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.20994","citing_title":"Towards Context-Invariant Safety Alignment for Large Language Models","ref_index":73,"is_internal_anchor":false},{"citing_arxiv_id":"2602.04583","citing_title":"PEPR: Privileged Event-based Predictive Regularization for Domain Generalization","ref_index":12,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CR3FIJISUIVV6A6MDPY4C634R5","json":"https://pith.science/pith/CR3FIJISUIVV6A6MDPY4C634R5.json","graph_json":"https://pith.science/api/pith-number/CR3FIJISUIVV6A6MDPY4C634R5/graph.json","events_json":"https://pith.science/api/pith-number/CR3FIJISUIVV6A6MDPY4C634R5/events.json","paper":"https://pith.science/paper/CR3FIJIS"},"agent_actions":{"view_html":"https://pith.science/pith/CR3FIJISUIVV6A6MDPY4C634R5","download_json":"https://pith.science/pith/CR3FIJISUIVV6A6MDPY4C634R5.json","view_paper":"https://pith.science/paper/CR3FIJIS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2004.05007&json=true","fetch_graph":"https://pith.science/api/pith-number/CR3FIJISUIVV6A6MDPY4C634R5/graph.json","fetch_events":"https://pith.science/api/pith-number/CR3FIJISUIVV6A6MDPY4C634R5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CR3FIJISUIVV6A6MDPY4C634R5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CR3FIJISUIVV6A6MDPY4C634R5/action/storage_attestation","attest_author":"https://pith.science/pith/CR3FIJISUIVV6A6MDPY4C634R5/action/author_attestation","sign_citation":"https://pith.science/pith/CR3FIJISUIVV6A6MDPY4C634R5/action/citation_signature","submit_replication":"https://pith.science/pith/CR3FIJISUIVV6A6MDPY4C634R5/action/replication_record"}},"created_at":"2026-07-05T01:16:14.484496+00:00","updated_at":"2026-07-05T01:16:14.484496+00:00"}