{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:2YMQDBWSPRAPB5UVNHGWHAWW6T","short_pith_number":"pith:2YMQDBWS","schema_version":"1.0","canonical_sha256":"d6190186d27c40f0f69569cd6382d6f4d02c607fa03e8f5e50af26ab1450c557","source":{"kind":"arxiv","id":"2202.06856","version":2},"attestation_state":"computed","paper":{"title":"Domain-Adjusted Regression or: ERM May Already Learn Features Sufficient for Out-of-Distribution Generalization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Andrej Risteski, Elan Rosenfeld, Pradeep Ravikumar","submitted_at":"2022-02-14T16:42:16Z","abstract_excerpt":"A common explanation for the failure of deep networks to generalize out-of-distribution is that they fail to recover the \"correct\" features. We challenge this notion with a simple experiment which suggests that ERM already learns sufficient features and that the current bottleneck is not feature learning, but robust regression. Our findings also imply that given a small amount of data from the target distribution, retraining only the last linear layer will give excellent performance. We therefore argue that devising simpler methods for learning predictors on existing features is a promising di"},"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":"2202.06856","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-02-14T16:42:16Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"b0a8720960ab12a212cf33d4eb7b3438d3d73baf8d7d4f03fa02e1450c32be1d","abstract_canon_sha256":"ac795059c2609edcab4a202dd9305573d4625fcecb85a532f13ed032cbc1111e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:11:18.733009Z","signature_b64":"vfJmxzLsFYY4yqag1h72RVv+Qb6w72nIfPLxsWTa6Qj5pxZ0GuKwglgFgDhzcBbj5fQi6JxZ7Qp7ghIMkZG6DQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d6190186d27c40f0f69569cd6382d6f4d02c607fa03e8f5e50af26ab1450c557","last_reissued_at":"2026-07-05T05:11:18.732631Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:11:18.732631Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Domain-Adjusted Regression or: ERM May Already Learn Features Sufficient for Out-of-Distribution Generalization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Andrej Risteski, Elan Rosenfeld, Pradeep Ravikumar","submitted_at":"2022-02-14T16:42:16Z","abstract_excerpt":"A common explanation for the failure of deep networks to generalize out-of-distribution is that they fail to recover the \"correct\" features. We challenge this notion with a simple experiment which suggests that ERM already learns sufficient features and that the current bottleneck is not feature learning, but robust regression. Our findings also imply that given a small amount of data from the target distribution, retraining only the last linear layer will give excellent performance. We therefore argue that devising simpler methods for learning predictors on existing features is a promising di"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2202.06856","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/2202.06856/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":"2202.06856","created_at":"2026-07-05T05:11:18.732686+00:00"},{"alias_kind":"arxiv_version","alias_value":"2202.06856v2","created_at":"2026-07-05T05:11:18.732686+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2202.06856","created_at":"2026-07-05T05:11:18.732686+00:00"},{"alias_kind":"pith_short_12","alias_value":"2YMQDBWSPRAP","created_at":"2026-07-05T05:11:18.732686+00:00"},{"alias_kind":"pith_short_16","alias_value":"2YMQDBWSPRAPB5UV","created_at":"2026-07-05T05:11:18.732686+00:00"},{"alias_kind":"pith_short_8","alias_value":"2YMQDBWS","created_at":"2026-07-05T05:11:18.732686+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2506.12226","citing_title":"Learning Causality for Modern Machine Learning","ref_index":53,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2YMQDBWSPRAPB5UVNHGWHAWW6T","json":"https://pith.science/pith/2YMQDBWSPRAPB5UVNHGWHAWW6T.json","graph_json":"https://pith.science/api/pith-number/2YMQDBWSPRAPB5UVNHGWHAWW6T/graph.json","events_json":"https://pith.science/api/pith-number/2YMQDBWSPRAPB5UVNHGWHAWW6T/events.json","paper":"https://pith.science/paper/2YMQDBWS"},"agent_actions":{"view_html":"https://pith.science/pith/2YMQDBWSPRAPB5UVNHGWHAWW6T","download_json":"https://pith.science/pith/2YMQDBWSPRAPB5UVNHGWHAWW6T.json","view_paper":"https://pith.science/paper/2YMQDBWS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2202.06856&json=true","fetch_graph":"https://pith.science/api/pith-number/2YMQDBWSPRAPB5UVNHGWHAWW6T/graph.json","fetch_events":"https://pith.science/api/pith-number/2YMQDBWSPRAPB5UVNHGWHAWW6T/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2YMQDBWSPRAPB5UVNHGWHAWW6T/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2YMQDBWSPRAPB5UVNHGWHAWW6T/action/storage_attestation","attest_author":"https://pith.science/pith/2YMQDBWSPRAPB5UVNHGWHAWW6T/action/author_attestation","sign_citation":"https://pith.science/pith/2YMQDBWSPRAPB5UVNHGWHAWW6T/action/citation_signature","submit_replication":"https://pith.science/pith/2YMQDBWSPRAPB5UVNHGWHAWW6T/action/replication_record"}},"created_at":"2026-07-05T05:11:18.732686+00:00","updated_at":"2026-07-05T05:11:18.732686+00:00"}