aiqbench analysis
This research introduces VERDICT, a plug-in framework that leverages fine-tuned Vision-Language Models to localize extraction evidence and calibrate confidence scores for document field extraction tasks. By utilizing open-weight VLMs instead of closed APIs, the system gains access to internal states that significantly improve AUROC metrics for both value and evidence confidence, even under distribution shifts with minimal labeling.
VERDICT: Visual Extraction-Result Decoder with Integrated Confidence and Traceable Evidence
arXiv:2610.23933v1 Announce Type: cross Abstract: Enterprises run schema-guided structured field extraction on documents with frontier multimodal models, and accuracy keeps improving as new models and systems arrive. Two things have not kept pace: the evidence for each extracted value, meaning the location of the document it was read from, and how far to trust the value and that evidence. Neither is a component of most extraction systems today, so we build them as a framework that attaches to an existing extractor and leaves it untouched. We present VERDICT, a plug-in framework that uses a small Vision--Language Model (VLM) to ground/localize the evidence once per document over a fixed external extractor's output, then calibrates value and evidence confidence. Fine-tuning the VLM for evidence localization substantially improves visual evidence quality and, more importantly, expands VERDICT's selective-prediction operating range. At a strict confidence threshold $0.80$, VERDICT with the fine-tuned VLM gives usable precision--coverage trade-offs for both value and evidence confidence across the in-distribution and out of distribution datasets, whereas the frozen setting often collapses to little or no coverage. Running an open-weight VLM for localization rather than a closed API makes its internal states available to the calibrator, and they are worth more than output log-probabilities: in-distribution AUROC rises from $0.730$ to $0.914$ for value confidence and from $0.681$ to $0.906$ for evidence confidence. This advantage does not survive a distribution shift. We show that the signal is displaced rather than destroyed, and that labeling roughly $120$ fields from the target dataset takes evidence confidence from $0.641$ to $0.905$ on one dataset and from $0.675$ to $0.829$ on the other.
Abstract:Enterprises run schema-guided structured field extraction on documents with frontier multimodal models, and accuracy keeps improving as new models and systems arrive. Two things have not kept pace: the evidence for each extracted value, meaning the location of the document it was read from, and how far to trust the value and that evidence. Neither is a component of most extraction systems today, so we build them as a framework that attaches to an existing extractor and leaves it untouched. We present VERDICT, a plug-in framework that uses a small Vision--Language Model (VLM) to ground/localize the evidence once per document over a fixed external extractor's output, then calibrates value and evidence confidence. Fine-tuning the VLM for evidence localization substantially improves visual evidence quality and, more importantly, expands VERDICT's selective-prediction operating range. At a strict confidence threshold $0.80$, VERDICT with the fine-tuned VLM gives usable precision--coverage trade-offs for both value and evidence confidence across the in-distribution and out of distribution datasets, whereas the frozen setting often collapses to little or no coverage. Running an open-weight VLM for localization rather than a closed API makes its internal states available to the calibrator, and they are worth more than output log-probabilities: in-distribution AUROC rises from $0.730$ to $0.914$ for value confidence and from $0.681$ to $0.906$ for evidence confidence. This advantage does not survive a distribution shift. We show that the signal is displaced rather than destroyed, and that labeling roughly $120$ fields from the target dataset takes evidence confidence from $0.641$ to $0.905$ on one dataset and from $0.675$ to $0.829$ on the other.
Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
arXiv:2610.23933 [cs.CV]
(or arXiv:2610.23933v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2610.23933 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 360 words