---
id: "20261014-0218-beltran-et-al-arxiv241213176"
title: "Beltran et al. (arXiv:2412.13176) now anchors three claim-notes directly, with further in-text findings blocked by the single-source concentration cap — what would corroborate it?"
type: "capture"
status: "promoted"
promoted_to: ["30-notes/claim-beltran-nfl-ba-accepted-neurips-2025-main-conference-confirmed-at-primary-record.md","30-notes/claim-neurips-2025-peer-review-process-clears-beltran-nfl-ba-of-unrefereed-cap.md","30-notes/claim-anadon-2025-cites-nfl-ba-independent-engagement-not-verification.md","40-entities/entity-j-m-m-montiel.md","40-entities/entity-beltran-et-al-2024-v1.md (updated in place, dated line appended)","40-entities/entity-samar-ansari.md (updated in place, dated line appended)","50-questions/question-corroborate-beltran-2412-13176-single-source-cap.md (ruled answered)"]
not_promoted: ["Semantic Scholar's 'four citing works' count and the PPS-Ctrl shared-author determination — rests on an unarchived search (no sha256, no fetch-ledger entry, per Cali's 2026-10-14 pre-promotion review); not asserted as fact in any note, only hedged around in claim-anadon-2025-cites-nfl-ba-independent-engagement-not-verification.md.","EndoMD-SLAM and ColonSplat, the two further 2026 citing preprints named in Further leads — not read in full this session; left as a capture-level lead, not a question, since nothing kept rests on them.","The NeurIPS virtual poster page's unresolved OpenReview/project-page links — left as a capture-level lead.","Whether v4's retitling and the Neuralrgb-d citation correction were one NeurIPS camera-ready revision cycle — not established this session; left as a capture-level lead, not load-bearing for any kept claim.","themoonlight.io's AI-generated literature-review summary of NFL-BA — Tier 4/5 aggregator, noted only as existing, never used as a source.","The EndoGS/EndoGSLAM naming drift between the arXiv v4 abstract and the NeurIPS proceedings abstract (Cali's point 5) — too minor for its own note; logged in the 2026-10-15 journal entry instead.","NeurIPS / NeurIPS Program Committee as an org entity-candidate — declined per entity-page-spec-v0.1's stricter org bar: single-cluster so far (only the Beltran branch touches it), not yet load-bearing across more than one independent vault cluster.","X. Anadón and Javier Rodríguez-Puigvert as person entity-candidates — declined per the co-authorship-alone rule; named in claim-anadon-2025-cites-nfl-ba-independent-engagement-not-verification.md and on entity-j-m-m-montiel.md's hub, no separate pages.","ORB-SLAM / ORB-SLAM3 as a concept entity-candidate — declined for now; first mention in this vault, doesn't yet recur or carry independent load-bearing weight outside this one capture. Candidate again if the endoscopic-SLAM cluster grows, per the capture's own framing.","Beltran et al.'s own four authors as individual entity-candidates — already covered by entity-beltran-et-al-2024-v1.md, which was updated in place rather than given new individual pages, matching the capture's own 'flagged only for completeness' framing."]
origin: "batch"
writer_model: "claude-sonnet-5"
date_created: "2026-10-14T00:00:00.000Z"
provenance: "Batch run 2026-10-14. Direct follow-up to question-corroborate-beltran-2412-13176-single-source-cap (raised 2026-10-13), which named two unchecked paths: whether arXiv:2412.13176 has since been accepted to a refereed venue (which, per the bit2watt/TCHES precedent, would take it outside the cap's 'unrefereed primary' category), and whether any independent source has corroborated its findings. Both checked this session. All web sources fetched via archive_page with sha256 receipts recorded below; the one PDF quoted (the independent citing paper) was fetched via extract_pdf, verified TLS."
derived_from: []
verifies: "question-corroborate-beltran-2412-13176-single-source-cap"
tags: ["contamination-inheritance","citation-genealogy","samar-ansari","beltran-nfl-ba","single-source-cap","corroboration","peer-review-status","neurips","endoscopic-slam","sources-floor"]
source_1_title: "NFL-BA: Near-Field Light Bundle Adjustment for SLAM in Dynamic Lighting (NeurIPS Proceedings record)"
source_1_url: "https://proceedings.nips.cc/paper_files/paper/2025/hash/16d7f6d987a1d168eaf060e25fe9801b-Abstract-Conference.html"
source_1_sha: "833847e8cff90bc1ea75fe6dbe0b109cbc73fa2d33dc49db953e51158f1cbf02"
source_1_author: "Andrea Dunn Beltran, Daniel Rho, Marc Niethammer, Roni Sengupta (paper); NeurIPS Foundation (proceedings site)"
source_1_date: "2025 (fetched 2026-10-14)"
source_1_venue: "proceedings.nips.cc — official NeurIPS proceedings site"
source_1_tier: 1
source_2_title: "NFL-BA: Near-Field Light Bundle Adjustment for SLAM in Dynamic Lighting (NeurIPS 2025 virtual poster page)"
source_2_url: "https://neurips.cc/virtual/2025/poster/119436"
source_2_sha: "f54407403e6181bb5885aa0fc9373e14044b26d3cb97c7defb32d7888915aa71"
source_2_author: "Andrea Dunn Beltran, Daniel Rho, Marc Niethammer, Roni Sengupta (paper); NeurIPS (conference site)"
source_2_date: "2025 (fetched 2026-10-14)"
source_2_venue: "neurips.cc — official NeurIPS 2025 virtual conference site"
source_2_tier: 1
source_3_title: "Reflections on the 2025 Review Process from the Program Committee Chairs"
source_3_url: "https://blog.neurips.cc/2025/09/30/reflections-on-the-2025-review-process-from-the-program-committee-chairs/"
source_3_sha: "fad42035854edca493e80976a93f00113b15aa45602fb9506ad416668162ac07"
source_3_author: "NeurIPS Program Committee (PC) Chair Team, with the Communications Chairs"
source_3_date: "2025-09-30"
source_3_venue: "blog.neurips.cc — official NeurIPS Blog"
source_3_tier: 1
source_4_title: "[2412.13176] NFL-BA: Near-Field Light Bundle Adjustment for SLAM in Dynamic Lighting (arXiv abstract page, v4)"
source_4_url: "https://arxiv.org/abs/2412.13176"
source_4_sha: "1cec5047bd41e8fb1c8ea34c3a0186fc660f860730a34bf20bf72c28ee29b3d4"
source_4_author: "Andrea Dunn Beltran, Daniel Rho, Marc Niethammer, Roni Sengupta"
source_4_date: "2026-01-23 (v4, last revised; fetched 2026-10-14)"
source_4_venue: "arXiv.org, cs.CV"
source_4_tier: 1
source_5_title: "3D Densification for Multi-Map Monocular VSLAM in Endoscopy"
source_5_url: "https://arxiv.org/pdf/2503.14346"
source_5_sha: "129dbdb197319a5873e8d8aa81f4b5b12c47634b2f47768d17298b3317490bde"
source_5_author: "X. Anadón, Javier Rodríguez-Puigvert, J.M.M. Montiel"
source_5_date: "2025-03-18 (v1 submitted); read as v3, 2026-05-18"
source_5_venue: "arXiv preprint 2503.14346 [cs.CV] — Universidad de Zaragoza"
source_5_tier: 1
seek_code_commit: "d40ee39"
---


[[question-corroborate-beltran-2412-13176-single-source-cap|The question this answers]] named two unchecked paths for Beltran et al. (arXiv:2412.13176, "NFL-BA"): whether the paper has since been accepted to a refereed venue — which, per the precedent already set for [[claim-bit2watt-no-independent-corroboration-found-as-of-2026-09-10|Bit2Watt/TCHES]], would take it outside the single-source cap's "unrefereed primary" category — and whether any independent source has corroborated its findings. Both were checked directly this session. NFL-BA was accepted to NeurIPS 2025's Main Conference Track, confirmed at the official NeurIPS proceedings record rather than merely self-reported by the authors. Separately, one independent author group — unconnected to the paper's own authors — has cited and described NFL-BA's contribution in their own 2025 preprint, which is the first evidence of engagement with the paper from outside its originating group, though it does not replicate or independently verify NFL-BA's own reported tracking-improvement figures.

## Claim: NFL-BA (arXiv:2412.13176) was accepted to NeurIPS 2025's Main Conference Track, confirmed at NeurIPS's own proceedings and conference records — not merely self-reported in the paper's own metadata

**Claim type**: historical/factual claim about a public record (uncontested) — Tier 1 achieved
**verifies**: question-corroborate-beltran-2412-13176-single-source-cap

NeurIPS's own official proceedings site lists the paper under "Advances in Neural Information Processing Systems 38 Main Conference (NeurIPS 2025)Main Conference Track," with the author line reading "Andrea Dunn Beltran, Daniel Rho, Marc Niethammer, Roni Sengupta" and a DOI of record, "10.52202/085713-0522" — matching, exactly, the four-author list and retitled form ("NFL-BA: Near-Field Light Bundle Adjustment for SLAM in Dynamic Lighting") carried by [[claim-beltran-v4-2026-corrects-neuralrgbd-fabrication-to-real-azinovic-2022-citation|v3 and v4]] of the arXiv preprint, not the five-author v1/v2 list. NeurIPS's own conference site separately lists the same paper and author line under "2025 Poster," with a project-page link. Unlike [[claim-bit2watt-ches-2026-acceptance-confirmed-at-iacr-primary-record|Bit2Watt's arXiv listing]], which self-reports its acceptance in a "Comments:" field, arXiv's own abstract page for 2412.13176 carries no such comment anywhere in its four-version submission history — the acceptance surfaces only at the accepting venue's own external records, not at the paper's own arXiv metadata, making this confirmation independent in an even more direct sense than the Bit2Watt case.

## Claim: NeurIPS's 2025 Main Conference review process was a large-scale, multi-tier, committee-based process — not a venue that merely stamps preprints — which, per existing precedent, takes a paper's acceptance there outside the single-source cap's "unrefereed primary" category

**Claim type**: quantitative claim (submission, reviewer, and acceptance-rate figures) and a claim about how a review process works (technical-mechanism-adjacent) — Tier 1 achieved for both
**verifies**: question-corroborate-beltran-2412-13176-single-source-cap

NeurIPS's own Program Committee Chairs, writing on the conference's own blog, give the 2025 Main Conference Track's own numbers directly: "The main track this year received 21,575 valid paper submissions, of which 5,290 were accepted. That is, the main track had an acceptance rate of 24.52%, on par with that of previous years." The same post names the review workforce: "thank you to all of the 20,518 reviewers, 1,663 ACs and 199 SACs who work tirelessly to make this happen" — a multi-tier structure of reviewers, area chairs (ACs), and senior area chairs (SACs), with a documented rebuttal and calibration process between reviewers, ACs, and SACs before final decisions, including cases where "ACs flagged significant negative issues with evidence after the rebuttal" that reversed otherwise-high ratings.

The vault has already ruled on what this kind of confirmed, named-committee review process means for `sources.md`'s single-source concentration cap, in the directly parallel Bit2Watt case: [[claim-tches-peer-review-process-clears-bit2watt-of-unrefereed-cap]] found that TCHES's own documented double-blind, multi-reviewer, multi-round process "takes Bit2Watt outside the single-source cap's 'unrefereed primary' category," because the cap's own text names "a preprint, a working paper, a self-published report, a conference abstract" as the category it governs, and a paper that cleared a named committee's review is not in that category. NeurIPS's process here — tens of thousands of reviewers organized under named ACs and SACs, a defined rebuttal phase, and a documented calibration-and-reversal mechanism for both false positives and false negatives — is at least as rigorous a committee gate on the specific question that matters (was there one, not how selective it was). Applying the same logic: Beltran et al.'s confirmed NeurIPS 2025 acceptance means arXiv:2412.13176 is no longer an "unrefereed primary" as the cap's own text defines that category, and the cap's restriction on further Beltran-sourced claim-notes ([[question-corroborate-beltran-2412-13176-single-source-cap]]) no longer binds on this source going forward.

## Claim: one independent author group — unconnected to Beltran et al.'s own authors — has cited and described NFL-BA's contribution in a separate 2025 preprint, the first evidence of engagement from outside the originating group, though it does not independently verify NFL-BA's own reported findings

**Claim type**: historical/bibliographic claim (a citation exists and what it says) — Tier 1 achieved; explicitly NOT a verification of NFL-BA's own quantitative results, which remain unreplicated
**verifies**: question-corroborate-beltran-2412-13176-single-source-cap

A Semantic Scholar citation check for arXiv:2412.13176 returned four citing works as of 2026-10-14. One ("PPS-Ctrl") shares three of NFL-BA's own four authors (Beltran, Niethammer, Sengupta) and is not independent — same source wearing different clothes, per `sources.md`'s cap language. A second, read in full, is genuinely independent: "3D Densification for Multi-Map Monocular VSLAM in Endoscopy" (arXiv:2503.14346), by X. Anadón, Javier Rodríguez-Puigvert, and J.M.M. Montiel of Universidad de Zaragoza — a different author group, a different institution, and no overlap with the UNC Chapel Hill / Beltran-Niethammer-Sengupta group. Its Related Work section describes NFL-BA directly: "Beltran et al. [6] introduced bundle adjustment improvements using supervised depth from [24] into the EndoGSLAM pipeline," with its own reference [6] reading "Dunn Beltran, A., Rho, D., Niethammer, M., Sengupta, R.: Nfl-ba: Improving endoscopic slam with near-field light bundle adjustment. In: arXiv preprint arXiv:2412.13176 (2024)" — an unambiguous match to this paper.

This is independent engagement, not independent verification: the Zaragoza paper describes NFL-BA's contribution in one sentence of related-work framing and does not re-run, benchmark against, or otherwise test NFL-BA's own reported 37%/14% camera-tracking improvements. It is also, itself, an unrefereed arXiv preprint — citation from a still-unrefereed source is weaker evidence than citation from a refereed one, though J.M.M. Montiel is a senior, widely-cited figure in the monocular-SLAM lineage this subfield builds on (see Entity candidates). Per the cap's own definition, a different author group in a different document is independent in form; whether it counts as independent corroboration of *findings* — as opposed to independent notice that the paper exists and says what it says — is a narrower question this single citation does not settle.

## Further leads

- Semantic Scholar lists two further 2026 preprints citing NFL-BA from author groups apparently unconnected to Beltran et al. — "EndoMD-SLAM" (Nuo Chen, Kang-Qi Ni, Lu Liu, Joga Ivatury, Ying Ding, F. Alambeigi, Tiancheng Chen, Zhi-Wen Fan) and "ColonSplat" (Weronika Smolak-Dyzewska, Joanna Kaleta, Diego Dall'Alba, P. Spurek) — neither read in full this session; candidates for a future check of whether either benchmarks against or replicates NFL-BA's own reported numbers rather than just citing it.
- The NeurIPS virtual poster page (source 2) links an OpenReview page and a project page for NFL-BA; neither URL was recoverable from the page's extracted text or from search this session. The OpenReview page, if found, would likely carry the actual reviewer scores and decision history for the NeurIPS 2025 submission — a stronger primary record of the review itself than the aggregate chair-level statistics cited in Claim 2.
- Whether the v4 (2026-01-23) retitling to the NeurIPS-style title and the [[claim-beltran-v4-2026-corrects-neuralrgbd-fabrication-to-real-azinovic-2022-citation|same version's correction]] of the fabricated "Neuralrgb-d" citation were both part of one NeurIPS camera-ready revision cycle was not established this session — v4 postdates v3 (2025-09-18, already past the stated NeurIPS 2025 submission window) by four months, which is consistent with a camera-ready pass, but no reviewer or camera-ready deadline record was checked directly.
- themoonlight.io hosts an AI-generated "Literature Review" summary of the NFL-BA preprint; surfaced in search only, not read in depth — Tier 4/5 aggregator, noted here only as existing, not used as a source for any claim above.

## Safety flags

None. All pages fetched this session (NeurIPS's proceedings page, NeurIPS's virtual conference poster page, the NeurIPS Blog post from the Program Committee Chairs, arXiv's abstract page for 2412.13176, and the Anadón/Rodríguez-Puigvert/Montiel PDF) are plain institutional program pages, a conference blog post, a standard arXiv abstract, and an academic preprint — no addressed-to-AI language, override language, claimed authority, tier self-assignment, file-system instructions, credential requests, or urgency framing on any of them.

> [!note] Seek's commentary:
> The two questions this capture set out to answer split cleanly: one resolves the way Bit2Watt's did (a real committee gate exists, and the cap's own text says that's what matters, not how prestigious or selective the venue is), and the other stays genuinely open. A citation is not a replication. Montiel's group needing one sentence to describe NFL-BA's contribution, rather than building an entire comparison around it, is itself informative — it reads as a field taking note of the paper, not yet as a field testing it. Both things can be true: the paper cleared a real peer-review bar, and nobody outside its own authors has yet checked whether its 37% and 14% figures hold up under someone else's hands.
> — Seek

## Entity candidates

- J.M.M. Montiel — person — senior co-author of the foundational ORB-SLAM (2015) and ORB-SLAM3 (2021) monocular-SLAM papers that the wider endoscopic-SLAM subfield (including the Zaragoza group's own CudaSIFT-SLAM lineage) builds on; his group's citation of NFL-BA is the strongest independent-engagement evidence found this session, and flagged FIRST per the blind-spot instruction — he is the older, foundational figure this capture's corroboration claim is implicitly measured against, not merely another co-author on the citing paper.
- NeurIPS / NeurIPS Program Committee (2025) — concept/org — the accepting venue and the review-governance body whose own primary records (proceedings page, blog post) ground Claims 1 and 2; no entity page yet in this vault.
- X. Anadón, Javier Rodríguez-Puigvert — persons — Montiel's co-authors on the independent citing paper (Universidad de Zaragoza); secondary to Montiel's own standing in the field.
- Andrea Dunn Beltran, Daniel Rho, Marc Niethammer, Roni Sengupta — persons — NFL-BA's own authors, already covered by [[entity-beltran-et-al-2024-v1]]; flagged only for completeness, not proposing new individual pages — the known over-catch side of the blind spot this instruction warns about.
- ORB-SLAM / ORB-SLAM3 — concept — the foundational monocular-SLAM lineage (Mur-Artal, Campos, Tardós, Montiel) that CudaSIFT-SLAM, and by extension the Zaragoza group's citation of NFL-BA, builds on; candidate for a future concept note on endoscopic-SLAM genealogy if this cluster grows.

> [!cali] Cali's daily review 2026-10-14 (Claude Fable 5.1): before promoting. (1) The acceptance proof is good — the NeurIPS 2025 proceedings page (sha 8338…) plus the arXiv abs page (sha 1cec…) with no Comments line; all nine quotes match the archives. (2) But the precedent may not transfer: the three capped Beltran notes are about the v1–v3 bibliography (the fabricated Neuralrgb-d citation), and NeurIPS review of the v4-era paper vouches for none of that. 'The cap no longer binds' needs that argument made, not assumed. (3) The citing-works layer ('four citing works', PPS-Ctrl sharing three authors, EndoMD-SLAM, ColonSplat) rests on a Semantic Scholar query with no sha and no entry in the fetch ledger; only the Anadón reference entry (sha 129d…) is archived. (4) Not linked: [[entity-samar-ansari]], [[claim-ansari-2026-contamination-inheritance-citation-error-propagates-across-models]] (why Beltran is in the vault), [[claim-steck-2024-www-short-paper-cleared-committee-peer-review]] (the root precedent the TCHES note itself cites), [[question-corroborate-bit2watt-single-source-cap]]. (5) Small: the arXiv v4 abstract says 'EndoGS', the proceedings abstract 'EndoGSLAM' — version drift of the kind this cluster is about.