{"paper":{"title":"Learning to Decipher from Pixels: A Case Study of Copiale","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"A neural model can map handwritten cipher images directly to plaintext without first transcribing the symbols.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Alicia Forn\\'es, Be\\'ata Megyesi, Giuseppe De Gregorio, Lei Kang, Raphaela Heil","submitted_at":"2026-04-26T12:51:41Z","abstract_excerpt":"Historical encrypted manuscripts require both paleographic interpretation of cipher symbols and cryptanalytic recovery of plaintext. Most existing computational workflows rely on a transcription-first paradigm, in which handwritten symbols are transcribed prior to decipherment. This intermediate step is labor-intensive, error-prone, and not always aligned with the goal of direct plaintext recovery. We propose an end-to-end, transcription-free approach that directly maps handwritten cipher images to plaintext. Using the Copiale cipher as a case study, we introduce the first text-line-level data"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Our results demonstrate that transcription-free image-to-plaintext decipherment is both feasible and effective for historical substitution ciphers, offering a simplified and scalable alternative to traditional pipelines.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The method assumes that a model pretrained on generic handwriting data can be effectively fine-tuned on a limited cipher-specific dataset to learn the direct visual-to-plaintext mapping without needing explicit symbol-level transcription or additional cryptanalytic constraints.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"An end-to-end neural network deciphers the Copiale cipher directly from line-level images to German plaintext without any transcription step, using pretraining on generic handwriting followed by cipher-specific fine-tuning.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A neural model can map handwritten cipher images directly to plaintext without first transcribing the symbols.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"9f2f6102cebd35b53c97f9ebfe2183c39b470183429e86e60a10fe78b32fd9a5"},"source":{"id":"2604.23683","kind":"arxiv","version":2},"verdict":{"id":"4b147998-f964-478a-86a9-75ca48ed0e3e","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-08T06:48:31.559552Z","strongest_claim":"Our results demonstrate that transcription-free image-to-plaintext decipherment is both feasible and effective for historical substitution ciphers, offering a simplified and scalable alternative to traditional pipelines.","one_line_summary":"An end-to-end neural network deciphers the Copiale cipher directly from line-level images to German plaintext without any transcription step, using pretraining on generic handwriting followed by cipher-specific fine-tuning.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The method assumes that a model pretrained on generic handwriting data can be effectively fine-tuned on a limited cipher-specific dataset to learn the direct visual-to-plaintext mapping without needing explicit symbol-level transcription or additional cryptanalytic constraints.","pith_extraction_headline":"A neural model can map handwritten cipher images directly to plaintext without first transcribing the symbols."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.23683/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-21T08:37:07.124891Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T22:53:53.094087Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"3fefd370b6dd974c1bb41fe620a8538e703e4ff5455e0938ff6d7b69cee917bc"},"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"}