Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2605.27091 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866911720460517376 |
|---|---|
| author | Hu, Anqi Wang, Zhiyuan Jia, Zijun Fu, Bo |
| author_facet | Hu, Anqi Wang, Zhiyuan Jia, Zijun Fu, Bo |
| contents | Reliable set-valued prediction provides a principled way to mitigate hallucinations in open-ended question answering (QA), yet existing conformal approaches typically rely on a fragile premise: finite sampling must already produce at least one admissible candidate, or calibration examples violating this condition are discarded. In this paper, we introduce MiRD, a two-stage framework that decomposes overall miscoverage into sampling failure and conditional selection failure. In Stage I, MiRD establishes an expectation-level marginal upper bound on the probability that finite sampling produces no admissible answer under a fixed budget. In Stage II, conditioned on sampling success, MiRD calibrates a conformal selection threshold using admission-correlated nonconformity scores defined over the full calibration set, thereby preserving calibration-set integrity. Across three open-ended QA datasets and eight models, MiRD controls sampling risk, conditional selection risk, and overall miscoverage, while yielding tighter first-stage bounds than PAC-style alternatives and more adaptive prediction sets than successful-only calibration. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_27091 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | MiRD: Reliable Set-Valued Prediction for Open-Ended Question Answering via Miscoverage Risk Decomposition Hu, Anqi Wang, Zhiyuan Jia, Zijun Fu, Bo Computation and Language Artificial Intelligence Reliable set-valued prediction provides a principled way to mitigate hallucinations in open-ended question answering (QA), yet existing conformal approaches typically rely on a fragile premise: finite sampling must already produce at least one admissible candidate, or calibration examples violating this condition are discarded. In this paper, we introduce MiRD, a two-stage framework that decomposes overall miscoverage into sampling failure and conditional selection failure. In Stage I, MiRD establishes an expectation-level marginal upper bound on the probability that finite sampling produces no admissible answer under a fixed budget. In Stage II, conditioned on sampling success, MiRD calibrates a conformal selection threshold using admission-correlated nonconformity scores defined over the full calibration set, thereby preserving calibration-set integrity. Across three open-ended QA datasets and eight models, MiRD controls sampling risk, conditional selection risk, and overall miscoverage, while yielding tighter first-stage bounds than PAC-style alternatives and more adaptive prediction sets than successful-only calibration. |
| title | MiRD: Reliable Set-Valued Prediction for Open-Ended Question Answering via Miscoverage Risk Decomposition |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2605.27091 |