{"paper":{"title":"Possibilistic Predictive Uncertainty for Deep Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Deep neural networks can quantify epistemic uncertainty by projecting possibilistic posteriors over parameters onto predictions via supremum operators and approximating them with learnable Dirichlet possibility functions.","cross_cats":["cs.AI","cs.CV"],"primary_cat":"cs.LG","authors_text":"Jeremie Houssineau, Piotr Koniusz, Yao Ni, Yew Soon Ong","submitted_at":"2026-05-01T12:14:01Z","abstract_excerpt":"Deep neural networks achieve impressive results across diverse applications, yet their overconfidence on unseen inputs necessitates reliable epistemic uncertainty modeling. Existing methods for uncertainty modeling face a fundamental dilemma: Bayesian approaches provide principled estimates but remain computationally prohibitive, while efficient second-order predictors lack rigorous connections between their specific objectives and epistemic uncertainty quantification. To resolve this dilemma, we introduce Dirichlet-approximated possibilistic posterior predictions (DAPPr), a principled framewo"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"we introduce Dirichlet-approximated possibilistic posterior predictions (DAPPr), a principled framework leveraging possibility theory. We define a possibilistic posterior over parameters, projects this posterior to the prediction space via supremum operators, and approximates the projected posterior using learnable Dirichlet possibility functions. This projection-and-approximation strategy yields a simple training objective with closed-form solutions.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the supremum-based projection of the possibilistic posterior onto prediction space, followed by Dirichlet approximation, rigorously quantifies epistemic uncertainty rather than merely producing a convenient training objective.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"DAPPr introduces a possibilistic framework that projects parameter posteriors to predictions via supremum and approximates them with Dirichlet possibility functions to yield efficient, closed-form epistemic uncertainty estimates.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Deep neural networks can quantify epistemic uncertainty by projecting possibilistic posteriors over parameters onto predictions via supremum operators and approximating them with learnable Dirichlet possibility functions.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"05d3d0c38df8ee8df7155688f2dee2e62f8c5b6adb693d3da8b1aa922e1a4045"},"source":{"id":"2605.00600","kind":"arxiv","version":2},"verdict":{"id":"c3f23cc6-b8a0-4724-bb31-31573d6cab5a","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-09T19:11:00.236027Z","strongest_claim":"we introduce Dirichlet-approximated possibilistic posterior predictions (DAPPr), a principled framework leveraging possibility theory. We define a possibilistic posterior over parameters, projects this posterior to the prediction space via supremum operators, and approximates the projected posterior using learnable Dirichlet possibility functions. This projection-and-approximation strategy yields a simple training objective with closed-form solutions.","one_line_summary":"DAPPr introduces a possibilistic framework that projects parameter posteriors to predictions via supremum and approximates them with Dirichlet possibility functions to yield efficient, closed-form epistemic uncertainty estimates.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the supremum-based projection of the possibilistic posterior onto prediction space, followed by Dirichlet approximation, rigorously quantifies epistemic uncertainty rather than merely producing a convenient training objective.","pith_extraction_headline":"Deep neural networks can quantify epistemic uncertainty by projecting possibilistic posteriors over parameters onto predictions via supremum operators and approximating them with learnable Dirichlet possibility functions."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.00600/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-20T19:39:11.131120Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T18:01:54.904392Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"c3e53a54b63b685833fd8c482361f5ef01d205a6acbb9027fe18dc197e978c1e"},"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"}