{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:4UFCUCV6M75RSPX2JGXVN2XBYB","short_pith_number":"pith:4UFCUCV6","schema_version":"1.0","canonical_sha256":"e50a2a0abe67fb193efa49af56eae1c05fb67cce145e304bf7f0c10e37a8ffb3","source":{"kind":"arxiv","id":"2208.14024","version":2},"attestation_state":"computed","paper":{"title":"Positive Difference Distribution for Image Outlier Detection using Normalizing Flows and Contrastive Data","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Christoph-Nikolas Straehle, Robert Schmier, Ullrich K\\\"othe","submitted_at":"2022-08-30T07:00:46Z","abstract_excerpt":"Detecting test data deviating from training data is a central problem for safe and robust machine learning. Likelihoods learned by a generative model, e.g., a normalizing flow via standard log-likelihood training, perform poorly as an outlier score. We propose to use an unlabelled auxiliary dataset and a probabilistic outlier score for outlier detection. We use a self-supervised feature extractor trained on the auxiliary dataset and train a normalizing flow on the extracted features by maximizing the likelihood on in-distribution data and minimizing the likelihood on the contrastive dataset. W"},"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":"2208.14024","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-08-30T07:00:46Z","cross_cats_sorted":[],"title_canon_sha256":"baeddb71e20c9a7f1fc548959c7ace87f9d5bcb6b33fcb264a3dde10d22334de","abstract_canon_sha256":"91a08c1eb7fa229b4f7628ebfa4ad4fc267f25255bd1b0f32ebf5a472460b1dc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:04:51.635830Z","signature_b64":"C3s/GqVkT5QUni5+ovTj7iNGr72pPUhp3APy7yrNd1uW0a9ew7kAq26aLF93mUllF0xQG8QbB7s5wt2qqKO+Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e50a2a0abe67fb193efa49af56eae1c05fb67cce145e304bf7f0c10e37a8ffb3","last_reissued_at":"2026-07-05T06:04:51.635404Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:04:51.635404Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Positive Difference Distribution for Image Outlier Detection using Normalizing Flows and Contrastive Data","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Christoph-Nikolas Straehle, Robert Schmier, Ullrich K\\\"othe","submitted_at":"2022-08-30T07:00:46Z","abstract_excerpt":"Detecting test data deviating from training data is a central problem for safe and robust machine learning. Likelihoods learned by a generative model, e.g., a normalizing flow via standard log-likelihood training, perform poorly as an outlier score. We propose to use an unlabelled auxiliary dataset and a probabilistic outlier score for outlier detection. We use a self-supervised feature extractor trained on the auxiliary dataset and train a normalizing flow on the extracted features by maximizing the likelihood on in-distribution data and minimizing the likelihood on the contrastive dataset. W"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2208.14024","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/2208.14024/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":"2208.14024","created_at":"2026-07-05T06:04:51.635461+00:00"},{"alias_kind":"arxiv_version","alias_value":"2208.14024v2","created_at":"2026-07-05T06:04:51.635461+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2208.14024","created_at":"2026-07-05T06:04:51.635461+00:00"},{"alias_kind":"pith_short_12","alias_value":"4UFCUCV6M75R","created_at":"2026-07-05T06:04:51.635461+00:00"},{"alias_kind":"pith_short_16","alias_value":"4UFCUCV6M75RSPX2","created_at":"2026-07-05T06:04:51.635461+00:00"},{"alias_kind":"pith_short_8","alias_value":"4UFCUCV6","created_at":"2026-07-05T06:04:51.635461+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.08709","citing_title":"Contrastive Normalizing Flows for Uncertainty-Aware Parameter Estimation","ref_index":18,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4UFCUCV6M75RSPX2JGXVN2XBYB","json":"https://pith.science/pith/4UFCUCV6M75RSPX2JGXVN2XBYB.json","graph_json":"https://pith.science/api/pith-number/4UFCUCV6M75RSPX2JGXVN2XBYB/graph.json","events_json":"https://pith.science/api/pith-number/4UFCUCV6M75RSPX2JGXVN2XBYB/events.json","paper":"https://pith.science/paper/4UFCUCV6"},"agent_actions":{"view_html":"https://pith.science/pith/4UFCUCV6M75RSPX2JGXVN2XBYB","download_json":"https://pith.science/pith/4UFCUCV6M75RSPX2JGXVN2XBYB.json","view_paper":"https://pith.science/paper/4UFCUCV6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2208.14024&json=true","fetch_graph":"https://pith.science/api/pith-number/4UFCUCV6M75RSPX2JGXVN2XBYB/graph.json","fetch_events":"https://pith.science/api/pith-number/4UFCUCV6M75RSPX2JGXVN2XBYB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4UFCUCV6M75RSPX2JGXVN2XBYB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4UFCUCV6M75RSPX2JGXVN2XBYB/action/storage_attestation","attest_author":"https://pith.science/pith/4UFCUCV6M75RSPX2JGXVN2XBYB/action/author_attestation","sign_citation":"https://pith.science/pith/4UFCUCV6M75RSPX2JGXVN2XBYB/action/citation_signature","submit_replication":"https://pith.science/pith/4UFCUCV6M75RSPX2JGXVN2XBYB/action/replication_record"}},"created_at":"2026-07-05T06:04:51.635461+00:00","updated_at":"2026-07-05T06:04:51.635461+00:00"}