{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:XVM6KXWE5UPPVSAHPESGTPPI7R","short_pith_number":"pith:XVM6KXWE","canonical_record":{"source":{"id":"2605.02439","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-04T10:37:09Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"eccd89ddae41b0152425282f54df7e9973c4e97699348a7d2220a31597aef6b0","abstract_canon_sha256":"dfefbcf4eda255b094b58e8af3e6bad2f435fba6129b3c6709243d61a77a7123"},"schema_version":"1.0"},"canonical_sha256":"bd59e55ec4ed1efac807792469bde8fc525d834c7cdc567175f2be418853a794","source":{"kind":"arxiv","id":"2605.02439","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.02439","created_at":"2026-05-20T00:04:33Z"},{"alias_kind":"arxiv_version","alias_value":"2605.02439v2","created_at":"2026-05-20T00:04:33Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.02439","created_at":"2026-05-20T00:04:33Z"},{"alias_kind":"pith_short_12","alias_value":"XVM6KXWE5UPP","created_at":"2026-05-20T00:04:33Z"},{"alias_kind":"pith_short_16","alias_value":"XVM6KXWE5UPPVSAH","created_at":"2026-05-20T00:04:33Z"},{"alias_kind":"pith_short_8","alias_value":"XVM6KXWE","created_at":"2026-05-20T00:04:33Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:XVM6KXWE5UPPVSAHPESGTPPI7R","target":"record","payload":{"canonical_record":{"source":{"id":"2605.02439","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-04T10:37:09Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"eccd89ddae41b0152425282f54df7e9973c4e97699348a7d2220a31597aef6b0","abstract_canon_sha256":"dfefbcf4eda255b094b58e8af3e6bad2f435fba6129b3c6709243d61a77a7123"},"schema_version":"1.0"},"canonical_sha256":"bd59e55ec4ed1efac807792469bde8fc525d834c7cdc567175f2be418853a794","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:04:33.664648Z","signature_b64":"NpbGs+BDNfUd75e6OPe3jNx/8Ni7aAeu2+lph8bD2Ngq0VARN13LJbGhDcLsW384SRW9pFpuNkLsPiA+2jltDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bd59e55ec4ed1efac807792469bde8fc525d834c7cdc567175f2be418853a794","last_reissued_at":"2026-05-20T00:04:33.663929Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:04:33.663929Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.02439","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:04:33Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"KLZCQYskC5sIFkutPn4KZ370/q+c3lHBKMk/ERk+S8SHDLu9WdvobyKogsnhCP4bhhrKq1YK/BYG/70hoHIjBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-07T17:20:44.164473Z"},"content_sha256":"1411a1f57a33e01e0e5d7613d8fbc29ab84aaeeec6d758e1c10baa3a2fadfc1e","schema_version":"1.0","event_id":"sha256:1411a1f57a33e01e0e5d7613d8fbc29ab84aaeeec6d758e1c10baa3a2fadfc1e"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:XVM6KXWE5UPPVSAHPESGTPPI7R","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Anomaly-Preference Image Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Reformulating anomaly image generation as preference learning allows diffusion models to create more realistic and diverse anomalous samples from limited data.","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Dan Wang, Fuyun Wang, Hui Yan, Sujia Huang, Tong Zhang, Xin Liu, Xu Guo, Yuanzhi Wang, Zhen Cui","submitted_at":"2026-05-04T10:37:09Z","abstract_excerpt":"Synthesizing realistic and diverse anomalous samples from limited data is vital for robust model generalization. However, existing methods struggle to reconcile fidelity and diversity, often hampered by distribution misalignment and overfitting, respectively.To mitigate this, we introduce Anomaly Preference Optimization,a novel paradigm that reformulates anomaly generation as a preference learning problem.Central to our approach is an implicit preference alignment mechanism that leverages real anomalies as positive references, deriving optimization signals directly from denoising trajectory de"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Extensive experiments demonstrate that [Anomaly Preference Optimization] significantly outperforms existing baselines, achieving state-of-the-art performance in both realism and diversity.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That an implicit preference alignment mechanism leveraging real anomalies as positive references can derive effective optimization signals directly from denoising trajectory deviations without distribution misalignment or overfitting, and that the Time-Aware Capacity Allocation module successfully reconciles fidelity and diversity.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Anomaly Preference Optimization reformulates anomalous image synthesis as preference learning with implicit alignment from real anomalies and a time-aware capacity allocation module for diffusion models to balance diversity and fidelity.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Reformulating anomaly image generation as preference learning allows diffusion models to create more realistic and diverse anomalous samples from limited data.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"c602ee956d646e17817f604cc5c99b48a8bc46121738312ea1cae8e05c95f0f0"},"source":{"id":"2605.02439","kind":"arxiv","version":2},"verdict":{"id":"5ccd2263-6a54-4b39-a315-f68645b23f16","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-08T18:40:54.630580Z","strongest_claim":"Extensive experiments demonstrate that [Anomaly Preference Optimization] significantly outperforms existing baselines, achieving state-of-the-art performance in both realism and diversity.","one_line_summary":"Anomaly Preference Optimization reformulates anomalous image synthesis as preference learning with implicit alignment from real anomalies and a time-aware capacity allocation module for diffusion models to balance diversity and fidelity.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That an implicit preference alignment mechanism leveraging real anomalies as positive references can derive effective optimization signals directly from denoising trajectory deviations without distribution misalignment or overfitting, and that the Time-Aware Capacity Allocation module successfully reconciles fidelity and diversity.","pith_extraction_headline":"Reformulating anomaly image generation as preference learning allows diffusion models to create more realistic and diverse anomalous samples from limited data."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.02439/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_compliance","ran_at":"2026-05-19T16:22:51.189120Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"05d329a97dd11ae54c4b56f135e2064d709bc2566956cce3737a3934c94b6303"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":3,"snapshot_sha256":"c858f8c9506fb44b21d994ccb6c6e2314d942814cfe8b959a3ec31e22d213847"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"5ccd2263-6a54-4b39-a315-f68645b23f16"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:04:33Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"BMJUI9Xh6r0IDDD6oTFyFZ9jpWd6SB4PCHbSmd3se8mvERyAJckIoeWEygsMHf7QBdgyZ6AVt1z10M57Ad8ICw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-07T17:20:44.165306Z"},"content_sha256":"30f1d84ad3cfa97778dd5719e3f24219532b0b9e45f6f41c61fb557e36c364e4","schema_version":"1.0","event_id":"sha256:30f1d84ad3cfa97778dd5719e3f24219532b0b9e45f6f41c61fb557e36c364e4"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/XVM6KXWE5UPPVSAHPESGTPPI7R/bundle.json","state_url":"https://pith.science/pith/XVM6KXWE5UPPVSAHPESGTPPI7R/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/XVM6KXWE5UPPVSAHPESGTPPI7R/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-06-07T17:20:44Z","links":{"resolver":"https://pith.science/pith/XVM6KXWE5UPPVSAHPESGTPPI7R","bundle":"https://pith.science/pith/XVM6KXWE5UPPVSAHPESGTPPI7R/bundle.json","state":"https://pith.science/pith/XVM6KXWE5UPPVSAHPESGTPPI7R/state.json","well_known_bundle":"https://pith.science/.well-known/pith/XVM6KXWE5UPPVSAHPESGTPPI7R/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:XVM6KXWE5UPPVSAHPESGTPPI7R","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":"dfefbcf4eda255b094b58e8af3e6bad2f435fba6129b3c6709243d61a77a7123","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-04T10:37:09Z","title_canon_sha256":"eccd89ddae41b0152425282f54df7e9973c4e97699348a7d2220a31597aef6b0"},"schema_version":"1.0","source":{"id":"2605.02439","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.02439","created_at":"2026-05-20T00:04:33Z"},{"alias_kind":"arxiv_version","alias_value":"2605.02439v2","created_at":"2026-05-20T00:04:33Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.02439","created_at":"2026-05-20T00:04:33Z"},{"alias_kind":"pith_short_12","alias_value":"XVM6KXWE5UPP","created_at":"2026-05-20T00:04:33Z"},{"alias_kind":"pith_short_16","alias_value":"XVM6KXWE5UPPVSAH","created_at":"2026-05-20T00:04:33Z"},{"alias_kind":"pith_short_8","alias_value":"XVM6KXWE","created_at":"2026-05-20T00:04:33Z"}],"graph_snapshots":[{"event_id":"sha256:30f1d84ad3cfa97778dd5719e3f24219532b0b9e45f6f41c61fb557e36c364e4","target":"graph","created_at":"2026-05-20T00:04:33Z","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":"Extensive experiments demonstrate that [Anomaly Preference Optimization] significantly outperforms existing baselines, achieving state-of-the-art performance in both realism and diversity."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That an implicit preference alignment mechanism leveraging real anomalies as positive references can derive effective optimization signals directly from denoising trajectory deviations without distribution misalignment or overfitting, and that the Time-Aware Capacity Allocation module successfully reconciles fidelity and diversity."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Anomaly Preference Optimization reformulates anomalous image synthesis as preference learning with implicit alignment from real anomalies and a time-aware capacity allocation module for diffusion models to balance diversity and fidelity."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Reformulating anomaly image generation as preference learning allows diffusion models to create more realistic and diverse anomalous samples from limited data."}],"snapshot_sha256":"c602ee956d646e17817f604cc5c99b48a8bc46121738312ea1cae8e05c95f0f0"},"formal_canon":{"evidence_count":3,"snapshot_sha256":"c858f8c9506fb44b21d994ccb6c6e2314d942814cfe8b959a3ec31e22d213847"},"integrity":{"available":true,"clean":true,"detectors_run":[{"findings_count":0,"name":"doi_compliance","ran_at":"2026-05-19T16:22:51.189120Z","status":"completed","version":"1.0.0"}],"endpoint":"/pith/2605.02439/integrity.json","findings":[],"snapshot_sha256":"05d329a97dd11ae54c4b56f135e2064d709bc2566956cce3737a3934c94b6303","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Synthesizing realistic and diverse anomalous samples from limited data is vital for robust model generalization. However, existing methods struggle to reconcile fidelity and diversity, often hampered by distribution misalignment and overfitting, respectively.To mitigate this, we introduce Anomaly Preference Optimization,a novel paradigm that reformulates anomaly generation as a preference learning problem.Central to our approach is an implicit preference alignment mechanism that leverages real anomalies as positive references, deriving optimization signals directly from denoising trajectory de","authors_text":"Dan Wang, Fuyun Wang, Hui Yan, Sujia Huang, Tong Zhang, Xin Liu, Xu Guo, Yuanzhi Wang, Zhen Cui","cross_cats":["cs.LG"],"headline":"Reformulating anomaly image generation as preference learning allows diffusion models to create more realistic and diverse anomalous samples from limited data.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-04T10:37:09Z","title":"Anomaly-Preference Image Generation"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.02439","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-08T18:40:54.630580Z","id":"5ccd2263-6a54-4b39-a315-f68645b23f16","model_set":{"reader":"grok-4.3"},"one_line_summary":"Anomaly Preference Optimization reformulates anomalous image synthesis as preference learning with implicit alignment from real anomalies and a time-aware capacity allocation module for diffusion models to balance diversity and fidelity.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Reformulating anomaly image generation as preference learning allows diffusion models to create more realistic and diverse anomalous samples from limited data.","strongest_claim":"Extensive experiments demonstrate that [Anomaly Preference Optimization] significantly outperforms existing baselines, achieving state-of-the-art performance in both realism and diversity.","weakest_assumption":"That an implicit preference alignment mechanism leveraging real anomalies as positive references can derive effective optimization signals directly from denoising trajectory deviations without distribution misalignment or overfitting, and that the Time-Aware Capacity Allocation module successfully reconciles fidelity and diversity."}},"verdict_id":"5ccd2263-6a54-4b39-a315-f68645b23f16"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:1411a1f57a33e01e0e5d7613d8fbc29ab84aaeeec6d758e1c10baa3a2fadfc1e","target":"record","created_at":"2026-05-20T00:04:33Z","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":"dfefbcf4eda255b094b58e8af3e6bad2f435fba6129b3c6709243d61a77a7123","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-04T10:37:09Z","title_canon_sha256":"eccd89ddae41b0152425282f54df7e9973c4e97699348a7d2220a31597aef6b0"},"schema_version":"1.0","source":{"id":"2605.02439","kind":"arxiv","version":2}},"canonical_sha256":"bd59e55ec4ed1efac807792469bde8fc525d834c7cdc567175f2be418853a794","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"bd59e55ec4ed1efac807792469bde8fc525d834c7cdc567175f2be418853a794","first_computed_at":"2026-05-20T00:04:33.663929Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-20T00:04:33.663929Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"NpbGs+BDNfUd75e6OPe3jNx/8Ni7aAeu2+lph8bD2Ngq0VARN13LJbGhDcLsW384SRW9pFpuNkLsPiA+2jltDw==","signature_status":"signed_v1","signed_at":"2026-05-20T00:04:33.664648Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.02439","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:1411a1f57a33e01e0e5d7613d8fbc29ab84aaeeec6d758e1c10baa3a2fadfc1e","sha256:30f1d84ad3cfa97778dd5719e3f24219532b0b9e45f6f41c61fb557e36c364e4"],"state_sha256":"36473ed3729ff1cb170f8831be5daedec128d5ce3daa3ee5008bc49835bfff17"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"WDfdlBwj3hzvlfe2CpQnBpYqcFEw68s+zJuPQrd8q/4zKitw5eS6Xx8hJBWEiBAWpgDYwgc3JAIkHkPEWYJlBA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-07T17:20:44.169563Z","bundle_sha256":"64884d994a20f4d799302d61e3587dde18c21c6489010f5006b9a077c485dc7a"}}