{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:MF7DHFCVGWLRNFSXNXE7LFB2UY","short_pith_number":"pith:MF7DHFCV","schema_version":"1.0","canonical_sha256":"617e33945535971696576dc9f5943aa6137b4bab4e0c8d4a4d65324a551894c2","source":{"kind":"arxiv","id":"2308.00566","version":2},"attestation_state":"computed","paper":{"title":"Stochastic positional embeddings improve masked image modeling","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Amir Bar, Amir Globerson, Assaf Shocher, Florian Bordes, Mahmoud Assran, Nicolas Ballas, Pascal Vincent, Trevor Darrell, Yann LeCun","submitted_at":"2023-07-31T17:59:08Z","abstract_excerpt":"Masked Image Modeling (MIM) is a promising self-supervised learning approach that enables learning from unlabeled images. Despite its recent success, learning good representations through MIM remains challenging because it requires predicting the right semantic content in accurate locations. For example, given an incomplete picture of a dog, we can guess that there is a tail, but we cannot determine its exact location. In this work, we propose to incorporate location uncertainty into MIM by using stochastic positional embeddings (StoP). Specifically, we condition the model on stochastic masked"},"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":"2308.00566","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2023-07-31T17:59:08Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"c2a524a97bf5e41f06bb379beb1e6bd72768b39a862dd7979ec07eed33c620ab","abstract_canon_sha256":"56f21c6c4f178a99a0b35d089df955f3d4aba62857ee71bb4fcf615ce82efb1b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:49:55.415870Z","signature_b64":"gWv9QbnAp1Chn4qTs+wvhx8FgTPFjDWwrpfep4ND3gwzwb8r/zrDi6noMJFuez+SZ7wEdEvjysj6l5ZI2CaiCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"617e33945535971696576dc9f5943aa6137b4bab4e0c8d4a4d65324a551894c2","last_reissued_at":"2026-07-05T07:49:55.415464Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:49:55.415464Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Stochastic positional embeddings improve masked image modeling","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Amir Bar, Amir Globerson, Assaf Shocher, Florian Bordes, Mahmoud Assran, Nicolas Ballas, Pascal Vincent, Trevor Darrell, Yann LeCun","submitted_at":"2023-07-31T17:59:08Z","abstract_excerpt":"Masked Image Modeling (MIM) is a promising self-supervised learning approach that enables learning from unlabeled images. Despite its recent success, learning good representations through MIM remains challenging because it requires predicting the right semantic content in accurate locations. For example, given an incomplete picture of a dog, we can guess that there is a tail, but we cannot determine its exact location. In this work, we propose to incorporate location uncertainty into MIM by using stochastic positional embeddings (StoP). Specifically, we condition the model on stochastic masked"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2308.00566","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/2308.00566/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":"2308.00566","created_at":"2026-07-05T07:49:55.415528+00:00"},{"alias_kind":"arxiv_version","alias_value":"2308.00566v2","created_at":"2026-07-05T07:49:55.415528+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2308.00566","created_at":"2026-07-05T07:49:55.415528+00:00"},{"alias_kind":"pith_short_12","alias_value":"MF7DHFCVGWLR","created_at":"2026-07-05T07:49:55.415528+00:00"},{"alias_kind":"pith_short_16","alias_value":"MF7DHFCVGWLRNFSX","created_at":"2026-07-05T07:49:55.415528+00:00"},{"alias_kind":"pith_short_8","alias_value":"MF7DHFCV","created_at":"2026-07-05T07:49:55.415528+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2511.08544","citing_title":"LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristics","ref_index":41,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MF7DHFCVGWLRNFSXNXE7LFB2UY","json":"https://pith.science/pith/MF7DHFCVGWLRNFSXNXE7LFB2UY.json","graph_json":"https://pith.science/api/pith-number/MF7DHFCVGWLRNFSXNXE7LFB2UY/graph.json","events_json":"https://pith.science/api/pith-number/MF7DHFCVGWLRNFSXNXE7LFB2UY/events.json","paper":"https://pith.science/paper/MF7DHFCV"},"agent_actions":{"view_html":"https://pith.science/pith/MF7DHFCVGWLRNFSXNXE7LFB2UY","download_json":"https://pith.science/pith/MF7DHFCVGWLRNFSXNXE7LFB2UY.json","view_paper":"https://pith.science/paper/MF7DHFCV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2308.00566&json=true","fetch_graph":"https://pith.science/api/pith-number/MF7DHFCVGWLRNFSXNXE7LFB2UY/graph.json","fetch_events":"https://pith.science/api/pith-number/MF7DHFCVGWLRNFSXNXE7LFB2UY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MF7DHFCVGWLRNFSXNXE7LFB2UY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MF7DHFCVGWLRNFSXNXE7LFB2UY/action/storage_attestation","attest_author":"https://pith.science/pith/MF7DHFCVGWLRNFSXNXE7LFB2UY/action/author_attestation","sign_citation":"https://pith.science/pith/MF7DHFCVGWLRNFSXNXE7LFB2UY/action/citation_signature","submit_replication":"https://pith.science/pith/MF7DHFCVGWLRNFSXNXE7LFB2UY/action/replication_record"}},"created_at":"2026-07-05T07:49:55.415528+00:00","updated_at":"2026-07-05T07:49:55.415528+00:00"}