{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:VBA26UM7XKZYXG2D4HXYQW6JDQ","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"112280043ddbcafb9f0b569126ba471a7d61a44638d7c5da1ea8ff3b70772e88","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T08:37:59Z","title_canon_sha256":"767685429e53469c55d11cbbbfb667cad6f9e6cc26032a48ac1876b5628df505"},"schema_version":"1.0","source":{"id":"2605.13175","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.13175","created_at":"2026-05-18T03:08:56Z"},{"alias_kind":"arxiv_version","alias_value":"2605.13175v1","created_at":"2026-05-18T03:08:56Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.13175","created_at":"2026-05-18T03:08:56Z"},{"alias_kind":"pith_short_12","alias_value":"VBA26UM7XKZY","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"VBA26UM7XKZYXG2D","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"VBA26UM7","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:1b6d6ffefcf76f63505fa11f8568aabb563bf1379763767cf96513191fa05f7d","target":"graph","created_at":"2026-05-18T03:08:56Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"We show that HT noise makes the statistical estimation problem harder, leading to less favorable sampling-error bounds."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the derived sampling-error bounds for the two representative diffusion models are tight enough to reflect practical performance differences between HT and LT noise across the tested regimes."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Heavy-tailed noise in diffusion models leads to less favorable sampling-error bounds than light-tailed Gaussian noise by making the underlying statistical estimation problem harder."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Heavy-tailed noise makes statistical estimation harder in diffusion models than Gaussian noise."}],"snapshot_sha256":"384378a8dde9f257125ea9755595684e37547bf170fffd90e0b5e784cc8c7c53"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"Recent works have proposed incorporating heavy-tailed (HT) noise into diffusion- and flow-based generative models, with the goals of better recovering the tails of target distributions and improving generative diversity. This motivation is intuitive: if the data are heavy-tailed, HT noise may appear better matched than light-tailed (LT) Gaussian noise. However, replacing Gaussian noise by HT noise also changes the underlying estimation problem. In this paper, we revisit this paradigm through a combined theoretical and empirical study, establishing sampling-error bounds for two representative d","authors_text":"Antonio Ocello, Hamza Cherkaoui, H\\'el\\`ene Halconruy","cross_cats":[],"headline":"Heavy-tailed noise makes statistical estimation harder in diffusion models than Gaussian noise.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T08:37:59Z","title":"Do Heavy Tails Help Diffusion? On the Subtle Trade-off Between Initialization and Training"},"references":{"count":15,"internal_anchors":2,"resolved_work":15,"sample":[{"cited_arxiv_id":"2603.00772","doi":"","is_internal_anchor":true,"ref_index":1,"title":"Initialization-aware score-based diffusion sampling.arXiv preprint arXiv:2603.00772,","work_id":"fe0b9c84-3c6b-4130-9088-13ad868bd075","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Diffusion generative models meet compressed sensing, with applications to imaging and finance.arXiv preprint arXiv:2509.03898,","work_id":"98b0239a-3092-41f2-9051-bef7217df386","year":null},{"cited_arxiv_id":"","doi":"10.1007/978-3-030-52915-4","is_internal_anchor":false,"ref_index":3,"title":"Springer Series in Operations Research and Financial Engineering","work_id":"1dc631c0-8b4f-44e0-8e51-1dd71862305c","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Heavy-tailed diffusion with denoising lévy probabilistic models","work_id":"f1716af4-56e8-4ead-a5ec-365690f50fac","year":2025},{"cited_arxiv_id":"2604.06065","doi":"","is_internal_anchor":true,"ref_index":5,"title":"Lipschitz regularity in Flow Matching and Diffusion Models: sharp sampling rates and functional inequalities","work_id":"4fb70106-c9d6-48e2-9be8-707e4273105e","year":null}],"snapshot_sha256":"63a28508f3e127f57b60a157be34b2d2de9ae603f5a066d37730c8ce629d938f"},"source":{"id":"2605.13175","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-14T19:07:39.988222Z","id":"be39c0f9-5f95-4946-97f3-6e8b52e47314","model_set":{"reader":"grok-4.3"},"one_line_summary":"Heavy-tailed noise in diffusion models leads to less favorable sampling-error bounds than light-tailed Gaussian noise by making the underlying statistical estimation problem harder.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Heavy-tailed noise makes statistical estimation harder in diffusion models than Gaussian noise.","strongest_claim":"We show that HT noise makes the statistical estimation problem harder, leading to less favorable sampling-error bounds.","weakest_assumption":"That the derived sampling-error bounds for the two representative diffusion models are tight enough to reflect practical performance differences between HT and LT noise across the tested regimes."}},"verdict_id":"be39c0f9-5f95-4946-97f3-6e8b52e47314"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:a403cb37f37b286941655e2c5f34e43fdedbc229b72384deb54c22730ea350be","target":"record","created_at":"2026-05-18T03:08:56Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"112280043ddbcafb9f0b569126ba471a7d61a44638d7c5da1ea8ff3b70772e88","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T08:37:59Z","title_canon_sha256":"767685429e53469c55d11cbbbfb667cad6f9e6cc26032a48ac1876b5628df505"},"schema_version":"1.0","source":{"id":"2605.13175","kind":"arxiv","version":1}},"canonical_sha256":"a841af519fbab38b9b43e1ef885bc91c0b29086f5669175c1ffe955d5d9952b7","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"a841af519fbab38b9b43e1ef885bc91c0b29086f5669175c1ffe955d5d9952b7","first_computed_at":"2026-05-18T03:08:56.485470Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T03:08:56.485470Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"61zkCB+x8lTBzltIF9nA57ASIPAfx216xYrRbvTZKCMiZuNN12XUhvQfS+KEGfyKDdfuebMTPtXhUo/zxfVPDg==","signature_status":"signed_v1","signed_at":"2026-05-18T03:08:56.486202Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.13175","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:a403cb37f37b286941655e2c5f34e43fdedbc229b72384deb54c22730ea350be","sha256:1b6d6ffefcf76f63505fa11f8568aabb563bf1379763767cf96513191fa05f7d"],"state_sha256":"28fb3aa2e0d2e5fc86fd2bf610d85b38792c15cd7f2cd28dff7edacbabb63ee2"}