When Identical Rows Disagree: From Benchmark Identifiability to Replication-Robust Anomaly Detection

Abstract

A released table is often treated as an i.i.d. sample, although its repeated rows may encode business frequency, repeated entities, joins, resampling, or extraction errors. We show that this ambiguity creates a hidden measurement layer with three consequences: feature-identical rows impose an attained evaluation ceiling, row-weighted AUROC is sensitive to replication, and row-trained detectors learn a multiplicity-size-biased law. An exact-row audit of all 690 OddBench datasets finds train-test overlap in 355, feature-identical label conflict in 147, and a test anomaly identical to a training normal in 137. Switching from row to support weighting changes AUROC by at least 0.05 on 50-61 datasets across four classical detector geometries. We introduce SCOUT (Support-Count Orthogonalized Unsupervised Testing), a factorized anomaly detector that separates replication-invariant support evidence from exposure-aware count evidence. Factorwise split-conformal calibration yields marginal false-positive-rate control, while the support channel is exactly invariant to arbitrary positive row replication. On 686 OddBench datasets and five seeds, support-only SCOUT is non-inferior to row-wise Isolation Forest in raw AUROC and improves replication-invariant AUROC. External normal-support evaluations track nominal false-positive levels, and four backbones remain exactly unchanged under controlled replication. Semi-synthetic interventions show that conditional count modeling helps materially only under strong rate heterogeneity. These results specify when multiplicity should be treated as signal, nuisance, or uninterpretable without additional information.

Using Temporal Replication and Partial Pooling to Improve Practical Identifiability in Presence-Only Poisson Point Process Models
Camille Mottier, Iris Lang, Justine Nicolas +1
arXiv21 Oct 2026PreprintCC BY 4.0Full textStatisticsarXiv
Community Detection from Aggregated Relational Data: Identifiability and Spectral Recovery
Andrew Davison, Owen G. Ward
arXiv2 Oct 2026PreprintCC BY 4.0Full textStatisticsarXiv
WFDroneBench: A Benchmark for Sensor Placement and Drone Routing for Wildfire Detection
Romain Puech, Joseph Ye, Danique de Moor +2
arXiv17 Sep 2026PreprintCC BY 4.0Full textOptimisation & controlarXiv
Where Is the Truth When AI Agents Disagree? Dependence-Aware Belief Propagation for Unsupervised Truth Discovery (Extended Version)
Laure Berti-Equille
arXiv3 Nov 2026PreprintCC BY 4.0Full textStatisticsarXiv
Example of exponential many collisions of identical hard disks in the plane
Hongda Qiu
arXiv16 Oct 2026PreprintarXiv licenceFull textDynamical systems Geometry & topologyarXiv

Cite

BibTeX RIS

BibTeX

@misc{deng2026when,
  title = {{When Identical Rows Disagree: From Benchmark Identifiability to Replication-Robust Anomaly Detection}},
  author = {Jie Deng},
  howpublished = {arXiv},
  year = {2026},
  month = sep,
  doi = {10.48550/arxiv.2609.29580},
  eprint = {2609.29580},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.29580},
}

RIS

TY  - UNPB
TI  - When Identical Rows Disagree: From Benchmark Identifiability to Replication-Robust Anomaly Detection
AU  - Jie Deng
PY  - 2026
DA  - 2026/09/01
PB  - arXiv
DO  - 10.48550/arxiv.2609.29580
UR  - https://arxiv.org/abs/2609.29580
AB  - A released table is often treated as an i.i.d. sample, although its repeated rows may encode business frequency, repeated entities, joins, resampling, or extraction errors. We show that this ambiguity creates a hidden measurement layer with three consequences: feature-identical rows impose an attained evaluation ceiling, row-weighted AUROC is sensitive to replication, and row-trained detectors learn a multiplicity-size-biased law. An exact-row audit of all 690 OddBench datasets finds train-test overlap in 355, feature-identical label conflict in 147, and a test anomaly identical to a training normal in 137. Switching from row to support weighting changes AUROC by at least 0.05 on 50-61 datasets across four classical detector geometries. We introduce SCOUT (Support-Count Orthogonalized Unsupervised Testing), a factorized anomaly detector that separates replication-invariant support evidence from exposure-aware count evidence. Factorwise split-conformal calibration yields marginal false-positive-rate control, while the support channel is exactly invariant to arbitrary positive row replication. On 686 OddBench datasets and five seeds, support-only SCOUT is non-inferior to row-wise Isolation Forest in raw AUROC and improves replication-invariant AUROC. External normal-support evaluations track nominal false-positive levels, and four backbones remain exactly unchanged under controlled replication. Semi-synthetic interventions show that conditional count modeling helps materially only under strong rate heterogeneity. These results specify when multiplicity should be treated as signal, nuisance, or uninterpretable without additional information.
ER  -

Abstract

Related

Cite