{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2HLSBLSYHUZVVTZCXHRRRMXB46","short_pith_number":"pith:2HLSBLSY","schema_version":"1.0","canonical_sha256":"d1d720ae583d335acf22b9e318b2e1e7be0ca4bdc4e81a153ff4d8f7f66e45cc","source":{"kind":"arxiv","id":"2508.21695","version":1},"attestation_state":"computed","paper":{"title":"Activation Subspaces for Out-of-Distribution Detection","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Bar{\\i}\\c{s} Z\\\"ong\\\"ur, Robin Hesse, Stefan Roth","submitted_at":"2025-08-29T15:03:36Z","abstract_excerpt":"To ensure the reliability of deep models in real-world applications, out-of-distribution (OOD) detection methods aim to distinguish samples close to the training distribution (in-distribution, ID) from those farther away (OOD). In this work, we propose a novel OOD detection method that utilizes singular value decomposition of the weight matrix of the classification head to decompose the model's activations into decisive and insignificant components, which contribute maximally, respectively minimally, to the final classifier output. We find that the subspace of insignificant components more eff"},"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":"2508.21695","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-08-29T15:03:36Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"86eb75287d2f08ebeac2e954af4f939d5e1329e3265548e8fddc765b8ef2cd04","abstract_canon_sha256":"8853d3f60bfc4d6952cb430b106509f5b9e7d60cb37d224a16dfc0d7b2ff5198"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:01:48.068924Z","signature_b64":"ARsOQJrSfyz8LV99P2na49dZemdVx73TdJ5inP7Tegw70GqV+Ie+GNyrmSFDQv9Po2cqEcvT0yL1PcNCK0vqAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d1d720ae583d335acf22b9e318b2e1e7be0ca4bdc4e81a153ff4d8f7f66e45cc","last_reissued_at":"2026-07-05T12:01:48.068440Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:01:48.068440Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Activation Subspaces for Out-of-Distribution Detection","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Bar{\\i}\\c{s} Z\\\"ong\\\"ur, Robin Hesse, Stefan Roth","submitted_at":"2025-08-29T15:03:36Z","abstract_excerpt":"To ensure the reliability of deep models in real-world applications, out-of-distribution (OOD) detection methods aim to distinguish samples close to the training distribution (in-distribution, ID) from those farther away (OOD). In this work, we propose a novel OOD detection method that utilizes singular value decomposition of the weight matrix of the classification head to decompose the model's activations into decisive and insignificant components, which contribute maximally, respectively minimally, to the final classifier output. We find that the subspace of insignificant components more eff"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.21695","kind":"arxiv","version":1},"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/2508.21695/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":"2508.21695","created_at":"2026-07-05T12:01:48.068498+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.21695v1","created_at":"2026-07-05T12:01:48.068498+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.21695","created_at":"2026-07-05T12:01:48.068498+00:00"},{"alias_kind":"pith_short_12","alias_value":"2HLSBLSYHUZV","created_at":"2026-07-05T12:01:48.068498+00:00"},{"alias_kind":"pith_short_16","alias_value":"2HLSBLSYHUZVVTZC","created_at":"2026-07-05T12:01:48.068498+00:00"},{"alias_kind":"pith_short_8","alias_value":"2HLSBLSY","created_at":"2026-07-05T12:01:48.068498+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2HLSBLSYHUZVVTZCXHRRRMXB46","json":"https://pith.science/pith/2HLSBLSYHUZVVTZCXHRRRMXB46.json","graph_json":"https://pith.science/api/pith-number/2HLSBLSYHUZVVTZCXHRRRMXB46/graph.json","events_json":"https://pith.science/api/pith-number/2HLSBLSYHUZVVTZCXHRRRMXB46/events.json","paper":"https://pith.science/paper/2HLSBLSY"},"agent_actions":{"view_html":"https://pith.science/pith/2HLSBLSYHUZVVTZCXHRRRMXB46","download_json":"https://pith.science/pith/2HLSBLSYHUZVVTZCXHRRRMXB46.json","view_paper":"https://pith.science/paper/2HLSBLSY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.21695&json=true","fetch_graph":"https://pith.science/api/pith-number/2HLSBLSYHUZVVTZCXHRRRMXB46/graph.json","fetch_events":"https://pith.science/api/pith-number/2HLSBLSYHUZVVTZCXHRRRMXB46/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2HLSBLSYHUZVVTZCXHRRRMXB46/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2HLSBLSYHUZVVTZCXHRRRMXB46/action/storage_attestation","attest_author":"https://pith.science/pith/2HLSBLSYHUZVVTZCXHRRRMXB46/action/author_attestation","sign_citation":"https://pith.science/pith/2HLSBLSYHUZVVTZCXHRRRMXB46/action/citation_signature","submit_replication":"https://pith.science/pith/2HLSBLSYHUZVVTZCXHRRRMXB46/action/replication_record"}},"created_at":"2026-07-05T12:01:48.068498+00:00","updated_at":"2026-07-05T12:01:48.068498+00:00"}