Saved in:
Bibliographic Details
Main Author: Ortiz, Jorge
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2601.00138
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908768520896512
author Ortiz, Jorge
author_facet Ortiz, Jorge
contents High-stakes deployment of vision-language models (VLMs) requires selective prediction, where systems abstain when uncertain rather than risk costly errors. We investigate whether confidence-based abstention provides reliable control over error rates in video question answering, and whether that control remains robust under distribution shift. Using NExT-QA and Gemini 2.0 Flash, we establish two findings. First, confidence thresholding provides mechanistic control in-distribution. Sweeping threshold epsilon produces smooth risk-coverage tradeoffs, reducing error rates f
format Preprint
id arxiv_https___arxiv_org_abs_2601_00138
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Explicit Abstention Knobs for Predictable Reliability in Video Question Answering
Ortiz, Jorge
Artificial Intelligence
Computer Vision and Pattern Recognition
High-stakes deployment of vision-language models (VLMs) requires selective prediction, where systems abstain when uncertain rather than risk costly errors. We investigate whether confidence-based abstention provides reliable control over error rates in video question answering, and whether that control remains robust under distribution shift. Using NExT-QA and Gemini 2.0 Flash, we establish two findings. First, confidence thresholding provides mechanistic control in-distribution. Sweeping threshold epsilon produces smooth risk-coverage tradeoffs, reducing error rates f
title Explicit Abstention Knobs for Predictable Reliability in Video Question Answering
topic Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2601.00138