{"paper":{"title":"AdaptNC: Adaptive Nonconformity Scores for Conformal Prediction under Distribution Shift","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"AdaptNC adapts both nonconformity scores and thresholds online to shrink prediction regions under distribution shifts while preserving coverage.","cross_cats":["cs.RO","cs.SY","eess.SY"],"primary_cat":"cs.LG","authors_text":"Aditya Singh, Rahul Mangharam, Renukanandan Tumu","submitted_at":"2026-02-02T04:41:35Z","abstract_excerpt":"Rigorous uncertainty quantification is essential for the safe deployment of autonomous systems in unconstrained environments. Conformal Prediction (CP) provides a distribution-free framework for this task, yet its standard formulations rely on exchangeability assumptions that are violated by the distribution shifts inherent in real-world robotics. Existing online CP methods maintain target coverage by adaptively scaling the conformal threshold, but typically employ a static nonconformity score function. We show that this fixed geometry leads to highly conservative, volume-inefficient predictio"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"AdaptNC significantly reduces prediction region volume compared to state-of-the-art threshold-only baselines while maintaining target coverage levels.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The adaptive reweighting of nonconformity scores combined with the replay buffer preserves the marginal coverage guarantees of conformal prediction during online transitions under distribution shift.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"AdaptNC jointly adapts nonconformity scores and thresholds in conformal prediction to shrink prediction region volumes under distribution shifts while preserving target coverage.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"AdaptNC adapts both nonconformity scores and thresholds online to shrink prediction regions under distribution shifts while preserving coverage.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"e4f570b63a146021d9d8397e547ef5058051a927d6733e21e90003daa72051f2"},"source":{"id":"2602.01629","kind":"arxiv","version":2},"verdict":{"id":"b28bc6b9-f206-4263-8a90-c387270d618c","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T08:04:01.606656Z","strongest_claim":"AdaptNC significantly reduces prediction region volume compared to state-of-the-art threshold-only baselines while maintaining target coverage levels.","one_line_summary":"AdaptNC jointly adapts nonconformity scores and thresholds in conformal prediction to shrink prediction region volumes under distribution shifts while preserving target coverage.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The adaptive reweighting of nonconformity scores combined with the replay buffer preserves the marginal coverage guarantees of conformal prediction during online transitions under distribution shift.","pith_extraction_headline":"AdaptNC adapts both nonconformity scores and thresholds online to shrink prediction regions under distribution shifts while preserving coverage."},"references":{"count":16,"sample":[{"doi":"10.1016/j.jmva.2005","year":2005,"title":"doi: 10.1016/j.jmva.2005","work_id":"50ca57c1-e2c3-4818-927c-f0ec18a943bc","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Gao, C., Shan, L., Srinivas, V ., and Vijayaragha- van, A","work_id":"b757e598-257e-41a0-b920-95020b3b5023","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2021,"title":"URL https://openreview.net/forum? id=oNDhnGrD51&noteId=7kR09SC5BY. Gibbs, I. and Candes, E. Adaptive Conformal In- ference Under Distribution Shift. InAdvances in Neural Information Processing Systems","work_id":"92fc93d0-606a-4ced-b0c7-05b8f7bafa25","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1103/physreve.51","year":1991,"title":"URL http://jmlr.org/ papers/v25/22-1218.html","work_id":"f21483a7-5973-4621-8e28-7cf1941b8a6d","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.52202/079017-3158","year":1994,"title":"1103/PhysRevE.51.4282","work_id":"c854b72c-6d0d-4ddb-9324-26d799cc2ea3","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":16,"snapshot_sha256":"9dabc616d6ee04390e156fe906be6285f2329273c6c7c636f364edbeb5386dde","internal_anchors":0},"formal_canon":{"evidence_count":1,"snapshot_sha256":"00f52e30f5c96ed755e84972232695e1a38d1d615ec4152fc4f263275f8334d2"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}