{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:32N4TVWOZLHN4IAXROK3CUAUY7","short_pith_number":"pith:32N4TVWO","schema_version":"1.0","canonical_sha256":"de9bc9d6cecacede20178b95b15014c7e7d41e40b1c2f6fc959802963494dc70","source":{"kind":"arxiv","id":"1904.13132","version":3},"attestation_state":"computed","paper":{"title":"A critical analysis of self-supervision, or what we can learn from a single image","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Andrea Vedaldi, Christian Rupprecht, Yuki M. Asano","submitted_at":"2019-04-30T10:10:38Z","abstract_excerpt":"We look critically at popular self-supervision techniques for learning deep convolutional neural networks without manual labels. We show that three different and representative methods, BiGAN, RotNet and DeepCluster, can learn the first few layers of a convolutional network from a single image as well as using millions of images and manual labels, provided that strong data augmentation is used. However, for deeper layers the gap with manual supervision cannot be closed even if millions of unlabelled images are used for training. We conclude that: (1) the weights of the early layers of deep net"},"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":"1904.13132","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2019-04-30T10:10:38Z","cross_cats_sorted":[],"title_canon_sha256":"02e6d24d9249b3d18ce8a9c21bcc774ec859ed95e0565735a81262f1098f075b","abstract_canon_sha256":"351f4618a6bf68d7d61288b387eb1d629c92f39335926b0b91a16fe2081907b6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:42:18.953754Z","signature_b64":"Abup2SQ/7vgx04B5ebpe9gx7/0yPHC3dQc6UGmEgP7w29jMNzuoJeqOvX9Ef2qNPVudM8Qg5PGxclLFOTyrTBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"de9bc9d6cecacede20178b95b15014c7e7d41e40b1c2f6fc959802963494dc70","last_reissued_at":"2026-07-05T00:42:18.953375Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:42:18.953375Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A critical analysis of self-supervision, or what we can learn from a single image","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Andrea Vedaldi, Christian Rupprecht, Yuki M. Asano","submitted_at":"2019-04-30T10:10:38Z","abstract_excerpt":"We look critically at popular self-supervision techniques for learning deep convolutional neural networks without manual labels. We show that three different and representative methods, BiGAN, RotNet and DeepCluster, can learn the first few layers of a convolutional network from a single image as well as using millions of images and manual labels, provided that strong data augmentation is used. However, for deeper layers the gap with manual supervision cannot be closed even if millions of unlabelled images are used for training. We conclude that: (1) the weights of the early layers of deep net"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1904.13132","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":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/1904.13132/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":"1904.13132","created_at":"2026-07-05T00:42:18.953434+00:00"},{"alias_kind":"arxiv_version","alias_value":"1904.13132v3","created_at":"2026-07-05T00:42:18.953434+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1904.13132","created_at":"2026-07-05T00:42:18.953434+00:00"},{"alias_kind":"pith_short_12","alias_value":"32N4TVWOZLHN","created_at":"2026-07-05T00:42:18.953434+00:00"},{"alias_kind":"pith_short_16","alias_value":"32N4TVWOZLHN4IAX","created_at":"2026-07-05T00:42:18.953434+00:00"},{"alias_kind":"pith_short_8","alias_value":"32N4TVWO","created_at":"2026-07-05T00:42:18.953434+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.07230","citing_title":"`Attention-Guided Cross-Temporal Clustering for Self-Supervised Video Object Segmentation","ref_index":56,"is_internal_anchor":true},{"citing_arxiv_id":"2002.05709","citing_title":"A Simple Framework for Contrastive Learning of Visual Representations","ref_index":1,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/32N4TVWOZLHN4IAXROK3CUAUY7","json":"https://pith.science/pith/32N4TVWOZLHN4IAXROK3CUAUY7.json","graph_json":"https://pith.science/api/pith-number/32N4TVWOZLHN4IAXROK3CUAUY7/graph.json","events_json":"https://pith.science/api/pith-number/32N4TVWOZLHN4IAXROK3CUAUY7/events.json","paper":"https://pith.science/paper/32N4TVWO"},"agent_actions":{"view_html":"https://pith.science/pith/32N4TVWOZLHN4IAXROK3CUAUY7","download_json":"https://pith.science/pith/32N4TVWOZLHN4IAXROK3CUAUY7.json","view_paper":"https://pith.science/paper/32N4TVWO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1904.13132&json=true","fetch_graph":"https://pith.science/api/pith-number/32N4TVWOZLHN4IAXROK3CUAUY7/graph.json","fetch_events":"https://pith.science/api/pith-number/32N4TVWOZLHN4IAXROK3CUAUY7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/32N4TVWOZLHN4IAXROK3CUAUY7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/32N4TVWOZLHN4IAXROK3CUAUY7/action/storage_attestation","attest_author":"https://pith.science/pith/32N4TVWOZLHN4IAXROK3CUAUY7/action/author_attestation","sign_citation":"https://pith.science/pith/32N4TVWOZLHN4IAXROK3CUAUY7/action/citation_signature","submit_replication":"https://pith.science/pith/32N4TVWOZLHN4IAXROK3CUAUY7/action/replication_record"}},"created_at":"2026-07-05T00:42:18.953434+00:00","updated_at":"2026-07-05T00:42:18.953434+00:00"}