{"paper":{"title":"Vision-Based Runtime Monitoring under Varying Specifications using Semantic Latent Representations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"The vector of atom robustness scores is the minimal reusable interface that certifies any past-time STL formula from images after one calibration.","cross_cats":["cs.CV","cs.RO","cs.SY","eess.SY"],"primary_cat":"cs.LG","authors_text":"Bardh Hoxha, Georgios Fainekos, Hideki Okamoto, Lars Lindemann, Oliver Sch\\\"on","submitted_at":"2026-05-13T14:22:25Z","abstract_excerpt":"We study certified runtime monitoring of past-time signal temporal logic (ptSTL) from visual observations under partial observability. The monitor must infer safety-relevant quantities from images and provide finite-sample guarantees, while being \\emph{reusable}: once trained and calibrated, it should certify any formula in a target fragment without per-formula retraining. For fragments induced by a finite dictionary of temporal atoms, we prove that the \\emph{semantic basis}, the vector of atom robustness scores, is the minimum prediction target within the class of monotone, 1-Lipschitz reusab"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"we prove that the semantic basis, the vector of atom robustness scores, is the minimum prediction target within the class of monotone, 1-Lipschitz reusable interfaces: any formula is evaluated by a deterministic decoder derived from the parse tree, and a single conformal calibration pass certifies the entire fragment with no union bound.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The assumption that the target fragment is induced by a finite dictionary of temporal atoms and that the interfaces are monotone and 1-Lipschitz; if these do not hold, the reusability and single-calibration guarantees may fail.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Semantic basis vectors of atom robustness scores enable reusable conformal-certified monitoring of ptSTL formulas from visual inputs, with a rolling alternative and validation on driving data.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"The vector of atom robustness scores is the minimal reusable interface that certifies any past-time STL formula from images after one calibration.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"8e98f6f30c11ec971e6fa44b7f1fdadb7558c1733d27a606df024954f933779b"},"source":{"id":"2605.13923","kind":"arxiv","version":1},"verdict":{"id":"43abc0ca-5072-4edd-b392-7767a175b91d","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T05:14:24.079399Z","strongest_claim":"we prove that the semantic basis, the vector of atom robustness scores, is the minimum prediction target within the class of monotone, 1-Lipschitz reusable interfaces: any formula is evaluated by a deterministic decoder derived from the parse tree, and a single conformal calibration pass certifies the entire fragment with no union bound.","one_line_summary":"Semantic basis vectors of atom robustness scores enable reusable conformal-certified monitoring of ptSTL formulas from visual inputs, with a rolling alternative and validation on driving data.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The assumption that the target fragment is induced by a finite dictionary of temporal atoms and that the interfaces are monotone and 1-Lipschitz; if these do not hold, the reusability and single-calibration guarantees may fail.","pith_extraction_headline":"The vector of atom robustness scores is the minimal reusable interface that certifies any past-time STL formula from images after one calibration."},"references":{"count":26,"sample":[{"doi":"","year":2023,"title":"Conformal prediction for STL runtime verification,","work_id":"6894dfe3-c34f-4ce9-8116-b8e2b820904e","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Conformal quantitative predictive monitoring of STL requirements for stochastic processes,","work_id":"83ac410a-bc2e-47a1-9d05-d1bce982c276","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Robust conformal prediction for STL runtime verification under distribution shift,","work_id":"577bfba7-9d44-4efd-81a8-1cc1114b021a","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2009,"title":"Robustness of temporal logic specifications for continuous-time signals,","work_id":"f88cc6fc-d672-4dd0-b91b-651c51bb95da","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2017,"title":"Robust online monitoring of signal temporal logic,","work_id":"17ac69de-7422-4c56-baa3-ceb0736cf2c5","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":26,"snapshot_sha256":"01ef5d31b2b6c2715a58718ae599461b369a9746669bc34a1141ba09776b1669","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"}