{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:WJ2VY2CEWJR3MZSOJI6WIHNAZL","short_pith_number":"pith:WJ2VY2CE","schema_version":"1.0","canonical_sha256":"b2755c6844b263b6664e4a3d641da0cae23e72e2212f5cd02e8d1ec1e9a26d8e","source":{"kind":"arxiv","id":"2305.13128","version":2},"attestation_state":"computed","paper":{"title":"GSURE-Based Diffusion Model Training with Corrupted Data","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","cs.LG"],"primary_cat":"eess.IV","authors_text":"Bahjat Kawar, Michael Elad, Noam Elata, Tomer Michaeli","submitted_at":"2023-05-22T15:27:20Z","abstract_excerpt":"Diffusion models have demonstrated impressive results in both data generation and downstream tasks such as inverse problems, text-based editing, classification, and more. However, training such models usually requires large amounts of clean signals which are often difficult or impossible to obtain. In this work, we propose a novel training technique for generative diffusion models based only on corrupted data. We introduce a loss function based on the Generalized Stein's Unbiased Risk Estimator (GSURE), and prove that under some conditions, it is equivalent to the training objective used in fu"},"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":"2305.13128","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"eess.IV","submitted_at":"2023-05-22T15:27:20Z","cross_cats_sorted":["cs.CV","cs.LG"],"title_canon_sha256":"03212a4e8e9480e1bf414c99355a7f50b6a9cdbac53dbfd9c7555c48550be413","abstract_canon_sha256":"5138f822e9487280678b910a88fde021ab20fbcc6a7631276d99acceff8e6890"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:31:40.018612Z","signature_b64":"93b6ivegID43L6JNPBNhdPFnplkEp8D6BXoaEXKVHmJmaVShocWwDd0qCL8pRW4/DGj91KUwEYNZ4PPsPqrXCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b2755c6844b263b6664e4a3d641da0cae23e72e2212f5cd02e8d1ec1e9a26d8e","last_reissued_at":"2026-07-05T08:31:40.018063Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:31:40.018063Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"GSURE-Based Diffusion Model Training with Corrupted Data","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","cs.LG"],"primary_cat":"eess.IV","authors_text":"Bahjat Kawar, Michael Elad, Noam Elata, Tomer Michaeli","submitted_at":"2023-05-22T15:27:20Z","abstract_excerpt":"Diffusion models have demonstrated impressive results in both data generation and downstream tasks such as inverse problems, text-based editing, classification, and more. However, training such models usually requires large amounts of clean signals which are often difficult or impossible to obtain. In this work, we propose a novel training technique for generative diffusion models based only on corrupted data. We introduce a loss function based on the Generalized Stein's Unbiased Risk Estimator (GSURE), and prove that under some conditions, it is equivalent to the training objective used in fu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.13128","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/2305.13128/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":"2305.13128","created_at":"2026-07-05T08:31:40.018124+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.13128v2","created_at":"2026-07-05T08:31:40.018124+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.13128","created_at":"2026-07-05T08:31:40.018124+00:00"},{"alias_kind":"pith_short_12","alias_value":"WJ2VY2CEWJR3","created_at":"2026-07-05T08:31:40.018124+00:00"},{"alias_kind":"pith_short_16","alias_value":"WJ2VY2CEWJR3MZSO","created_at":"2026-07-05T08:31:40.018124+00:00"},{"alias_kind":"pith_short_8","alias_value":"WJ2VY2CE","created_at":"2026-07-05T08:31:40.018124+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.12365","citing_title":"Ambient Diffusion Policy: Imitation Learning from Suboptimal Data in Robotics","ref_index":43,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06861","citing_title":"Christoffel-DPS: Optimal sensor placement in diffusion posterior sampling for arbitrary distributions","ref_index":54,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WJ2VY2CEWJR3MZSOJI6WIHNAZL","json":"https://pith.science/pith/WJ2VY2CEWJR3MZSOJI6WIHNAZL.json","graph_json":"https://pith.science/api/pith-number/WJ2VY2CEWJR3MZSOJI6WIHNAZL/graph.json","events_json":"https://pith.science/api/pith-number/WJ2VY2CEWJR3MZSOJI6WIHNAZL/events.json","paper":"https://pith.science/paper/WJ2VY2CE"},"agent_actions":{"view_html":"https://pith.science/pith/WJ2VY2CEWJR3MZSOJI6WIHNAZL","download_json":"https://pith.science/pith/WJ2VY2CEWJR3MZSOJI6WIHNAZL.json","view_paper":"https://pith.science/paper/WJ2VY2CE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.13128&json=true","fetch_graph":"https://pith.science/api/pith-number/WJ2VY2CEWJR3MZSOJI6WIHNAZL/graph.json","fetch_events":"https://pith.science/api/pith-number/WJ2VY2CEWJR3MZSOJI6WIHNAZL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WJ2VY2CEWJR3MZSOJI6WIHNAZL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WJ2VY2CEWJR3MZSOJI6WIHNAZL/action/storage_attestation","attest_author":"https://pith.science/pith/WJ2VY2CEWJR3MZSOJI6WIHNAZL/action/author_attestation","sign_citation":"https://pith.science/pith/WJ2VY2CEWJR3MZSOJI6WIHNAZL/action/citation_signature","submit_replication":"https://pith.science/pith/WJ2VY2CEWJR3MZSOJI6WIHNAZL/action/replication_record"}},"created_at":"2026-07-05T08:31:40.018124+00:00","updated_at":"2026-07-05T08:31:40.018124+00:00"}