{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2018:YMXIHOJUDACASRKIVJSHS27DJ3","short_pith_number":"pith:YMXIHOJU","schema_version":"1.0","canonical_sha256":"c32e83b9341804094548aa64796be34ef8c4a9a33af8103b24742ae9be2520c6","source":{"kind":"arxiv","id":"1810.09136","version":3},"attestation_state":"computed","paper":{"title":"Do Deep Generative Models Know What They Don't Know?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Akihiro Matsukawa, Balaji Lakshminarayanan, Dilan Gorur, Eric Nalisnick, Yee Whye Teh","submitted_at":"2018-10-22T08:32:02Z","abstract_excerpt":"A neural network deployed in the wild may be asked to make predictions for inputs that were drawn from a different distribution than that of the training data. A plethora of work has demonstrated that it is easy to find or synthesize inputs for which a neural network is highly confident yet wrong. Generative models are widely viewed to be robust to such mistaken confidence as modeling the density of the input features can be used to detect novel, out-of-distribution inputs. In this paper we challenge this assumption. We find that the density learned by flow-based models, VAEs, and PixelCNNs ca"},"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":"1810.09136","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2018-10-22T08:32:02Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"7e90e7bb153cdf9e73cea26b67dac8d341828f83d2e9be091bda7e52bbc7138b","abstract_canon_sha256":"0a0f80a05849d9e162ddeaa0c1912c7fe2e35643dc9b5638d3ba9f4e02b1da49"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:52:51.825471Z","signature_b64":"xtM92zJBboi3UnVT/wkE8C/ybEwxo5pYH/dJv+cjdQ8LXZWvrAGE4cU0dETwtwNLKczXJCYxX1x8dhW7s8hzCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c32e83b9341804094548aa64796be34ef8c4a9a33af8103b24742ae9be2520c6","last_reissued_at":"2026-05-17T23:52:51.824755Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:52:51.824755Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Do Deep Generative Models Know What They Don't Know?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Akihiro Matsukawa, Balaji Lakshminarayanan, Dilan Gorur, Eric Nalisnick, Yee Whye Teh","submitted_at":"2018-10-22T08:32:02Z","abstract_excerpt":"A neural network deployed in the wild may be asked to make predictions for inputs that were drawn from a different distribution than that of the training data. A plethora of work has demonstrated that it is easy to find or synthesize inputs for which a neural network is highly confident yet wrong. Generative models are widely viewed to be robust to such mistaken confidence as modeling the density of the input features can be used to detect novel, out-of-distribution inputs. In this paper we challenge this assumption. We find that the density learned by flow-based models, VAEs, and PixelCNNs ca"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1810.09136","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":""},"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":"1810.09136","created_at":"2026-05-17T23:52:51.824880+00:00"},{"alias_kind":"arxiv_version","alias_value":"1810.09136v3","created_at":"2026-05-17T23:52:51.824880+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1810.09136","created_at":"2026-05-17T23:52:51.824880+00:00"},{"alias_kind":"pith_short_12","alias_value":"YMXIHOJUDACA","created_at":"2026-05-18T12:33:04.347982+00:00"},{"alias_kind":"pith_short_16","alias_value":"YMXIHOJUDACASRKI","created_at":"2026-05-18T12:33:04.347982+00:00"},{"alias_kind":"pith_short_8","alias_value":"YMXIHOJU","created_at":"2026-05-18T12:33:04.347982+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":12,"internal_anchor_count":5,"sample":[{"citing_arxiv_id":"1907.04572","citing_title":"Out-of-Distribution Detection Using Neural Rendering Generative Models","ref_index":11,"is_internal_anchor":true},{"citing_arxiv_id":"2411.05183","citing_title":"Why CNN Features Are not Gaussian: A Statistical Anatomy of Deep Representations","ref_index":51,"is_internal_anchor":true},{"citing_arxiv_id":"2605.18472","citing_title":"Flowing with Confidence","ref_index":11,"is_internal_anchor":true},{"citing_arxiv_id":"2510.17381","citing_title":"Beyond Binary Out-of-Distribution Detection: Characterizing Distributional Shifts with Multi-Statistic Diffusion Trajectories","ref_index":14,"is_internal_anchor":true},{"citing_arxiv_id":"2605.13225","citing_title":"Mix, Don't Tune: Bilingual Pre-Training Outperforms Hyperparameter Search in Data-Constrained Settings","ref_index":16,"is_internal_anchor":true},{"citing_arxiv_id":"2605.08618","citing_title":"Beyond Toy Benchmarks: A Systematic Evaluation of OOD Detection Methods For Plant Pathology Classification","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08202","citing_title":"Beyond Penalization: Diffusion-based Out-of-Distribution Detection and Selective Regularization in Offline Reinforcement Learning","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05638","citing_title":"Scaling Pretrained Representations Enables Label-Free Out-of-Distribution Detection Without Fine-Tuning","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2209.14988","citing_title":"DreamFusion: Text-to-3D using 2D Diffusion","ref_index":126,"is_internal_anchor":false},{"citing_arxiv_id":"2209.14988","citing_title":"DreamFusion: Text-to-3D using 2D Diffusion","ref_index":83,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07193","citing_title":"Coupling Models for One-Step Discrete Generation","ref_index":67,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18804","citing_title":"Geometric Decoupling: Diagnosing the Structural Instability of Latent","ref_index":18,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YMXIHOJUDACASRKIVJSHS27DJ3","json":"https://pith.science/pith/YMXIHOJUDACASRKIVJSHS27DJ3.json","graph_json":"https://pith.science/api/pith-number/YMXIHOJUDACASRKIVJSHS27DJ3/graph.json","events_json":"https://pith.science/api/pith-number/YMXIHOJUDACASRKIVJSHS27DJ3/events.json","paper":"https://pith.science/paper/YMXIHOJU"},"agent_actions":{"view_html":"https://pith.science/pith/YMXIHOJUDACASRKIVJSHS27DJ3","download_json":"https://pith.science/pith/YMXIHOJUDACASRKIVJSHS27DJ3.json","view_paper":"https://pith.science/paper/YMXIHOJU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1810.09136&json=true","fetch_graph":"https://pith.science/api/pith-number/YMXIHOJUDACASRKIVJSHS27DJ3/graph.json","fetch_events":"https://pith.science/api/pith-number/YMXIHOJUDACASRKIVJSHS27DJ3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YMXIHOJUDACASRKIVJSHS27DJ3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YMXIHOJUDACASRKIVJSHS27DJ3/action/storage_attestation","attest_author":"https://pith.science/pith/YMXIHOJUDACASRKIVJSHS27DJ3/action/author_attestation","sign_citation":"https://pith.science/pith/YMXIHOJUDACASRKIVJSHS27DJ3/action/citation_signature","submit_replication":"https://pith.science/pith/YMXIHOJUDACASRKIVJSHS27DJ3/action/replication_record"}},"created_at":"2026-05-17T23:52:51.824880+00:00","updated_at":"2026-05-17T23:52:51.824880+00:00"}