{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:TXZKUBD743YWHC6VRFIO5HC6GP","short_pith_number":"pith:TXZKUBD7","schema_version":"1.0","canonical_sha256":"9df2aa047fe6f1638bd58950ee9c5e33f3fbf56e574cc5531023b2b5d516b697","source":{"kind":"arxiv","id":"2003.01826","version":1},"attestation_state":"computed","paper":{"title":"Watch your Up-Convolution: CNN Based Generative Deep Neural Networks are Failing to Reproduce Spectral Distributions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["eess.IV"],"primary_cat":"cs.CV","authors_text":"Janis Keuper, Margret Keuper, Ricard Durall","submitted_at":"2020-03-03T23:04:33Z","abstract_excerpt":"Generative convolutional deep neural networks, e.g. popular GAN architectures, are relying on convolution based up-sampling methods to produce non-scalar outputs like images or video sequences. In this paper, we show that common up-sampling methods, i.e. known as up-convolution or transposed convolution, are causing the inability of such models to reproduce spectral distributions of natural training data correctly. This effect is independent of the underlying architecture and we show that it can be used to easily detect generated data like deepfakes with up to 100% accuracy on public benchmark"},"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":"2003.01826","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2020-03-03T23:04:33Z","cross_cats_sorted":["eess.IV"],"title_canon_sha256":"a7004b68f3ad41219f5c7d9f2f13ed4a46db9508130a46b4acc67e133dcf3e7f","abstract_canon_sha256":"914f99d0f9a3c82516fba0e35ef567c274ddbfd55294e653fa6b2acd3bac2af0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:45:36.404074Z","signature_b64":"f9TQ/2T52KGdzeGFUKp+prZh8BANwA5DF8PZK5Qo1aUt/uXkjR5270EZwi9054p/kemR/Qy0f8ZLdZob1jV6BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9df2aa047fe6f1638bd58950ee9c5e33f3fbf56e574cc5531023b2b5d516b697","last_reissued_at":"2026-07-05T00:45:36.403651Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:45:36.403651Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Watch your Up-Convolution: CNN Based Generative Deep Neural Networks are Failing to Reproduce Spectral Distributions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["eess.IV"],"primary_cat":"cs.CV","authors_text":"Janis Keuper, Margret Keuper, Ricard Durall","submitted_at":"2020-03-03T23:04:33Z","abstract_excerpt":"Generative convolutional deep neural networks, e.g. popular GAN architectures, are relying on convolution based up-sampling methods to produce non-scalar outputs like images or video sequences. In this paper, we show that common up-sampling methods, i.e. known as up-convolution or transposed convolution, are causing the inability of such models to reproduce spectral distributions of natural training data correctly. This effect is independent of the underlying architecture and we show that it can be used to easily detect generated data like deepfakes with up to 100% accuracy on public benchmark"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2003.01826","kind":"arxiv","version":1},"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/2003.01826/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":"2003.01826","created_at":"2026-07-05T00:45:36.403717+00:00"},{"alias_kind":"arxiv_version","alias_value":"2003.01826v1","created_at":"2026-07-05T00:45:36.403717+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2003.01826","created_at":"2026-07-05T00:45:36.403717+00:00"},{"alias_kind":"pith_short_12","alias_value":"TXZKUBD743YW","created_at":"2026-07-05T00:45:36.403717+00:00"},{"alias_kind":"pith_short_16","alias_value":"TXZKUBD743YWHC6V","created_at":"2026-07-05T00:45:36.403717+00:00"},{"alias_kind":"pith_short_8","alias_value":"TXZKUBD7","created_at":"2026-07-05T00:45:36.403717+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2507.10236","citing_title":"Navigating the Challenges of AI-Generated Image Detection in the Wild: What Truly Matters?","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08226","citing_title":"SPECTRA-Net: Scalable Pipeline for Explainable Cross-domain Tensor Representations for AI-generated Images Detection","ref_index":6,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TXZKUBD743YWHC6VRFIO5HC6GP","json":"https://pith.science/pith/TXZKUBD743YWHC6VRFIO5HC6GP.json","graph_json":"https://pith.science/api/pith-number/TXZKUBD743YWHC6VRFIO5HC6GP/graph.json","events_json":"https://pith.science/api/pith-number/TXZKUBD743YWHC6VRFIO5HC6GP/events.json","paper":"https://pith.science/paper/TXZKUBD7"},"agent_actions":{"view_html":"https://pith.science/pith/TXZKUBD743YWHC6VRFIO5HC6GP","download_json":"https://pith.science/pith/TXZKUBD743YWHC6VRFIO5HC6GP.json","view_paper":"https://pith.science/paper/TXZKUBD7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2003.01826&json=true","fetch_graph":"https://pith.science/api/pith-number/TXZKUBD743YWHC6VRFIO5HC6GP/graph.json","fetch_events":"https://pith.science/api/pith-number/TXZKUBD743YWHC6VRFIO5HC6GP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TXZKUBD743YWHC6VRFIO5HC6GP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TXZKUBD743YWHC6VRFIO5HC6GP/action/storage_attestation","attest_author":"https://pith.science/pith/TXZKUBD743YWHC6VRFIO5HC6GP/action/author_attestation","sign_citation":"https://pith.science/pith/TXZKUBD743YWHC6VRFIO5HC6GP/action/citation_signature","submit_replication":"https://pith.science/pith/TXZKUBD743YWHC6VRFIO5HC6GP/action/replication_record"}},"created_at":"2026-07-05T00:45:36.403717+00:00","updated_at":"2026-07-05T00:45:36.403717+00:00"}