@inproceedings{cao-etal-2026-correlation,
title = "Correlation-Flip Visual Injection for Hallucination Mitigation in Large Vision-Language Models",
author = "Cao, Jiaxuan and
Shi, Zeyu and
Li, Longhua and
Wang, Lei and
Qi, Lei",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.1161/",
pages = "26585--26596",
ISBN = "979-8-89176-459-0",
abstract = "Large vision-language models (LVLMs) often hallucinate due to over-reliance on language priors. Existing methods implicitly assume that visual tokens maintain consistent semantics across layers and thus indiscriminately inject shallow-layer visual information into deeper layers. We identify a new phenomenon, termed correlation flip, where a visual token positively correlated with the decoding context in an intermediate layer becomes negatively correlated in the final layer, indicating that correct visual evidence is suppressed before prediction. Based on this finding, we propose Correlation-Flip Visual Injection (CFVI), a simple and training-free decoding-time method that selectively identifies correlation-flip tokens and injects corrective visual signals into the final-layer hidden state to restore suppressed evidence and mitigate hallucinations. CFVI can be directly applied to existing LVLMs, and introduces minimal computational overhead. Experiments on multiple benchmarks demonstrate that CFVI achieves strong performance while preserving language fluency."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="cao-etal-2026-correlation">
<titleInfo>
<title>Correlation-Flip Visual Injection for Hallucination Mitigation in Large Vision-Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jiaxuan</namePart>
<namePart type="family">Cao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zeyu</namePart>
<namePart type="family">Shi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Longhua</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lei</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lei</namePart>
<namePart type="family">Qi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Large vision-language models (LVLMs) often hallucinate due to over-reliance on language priors. Existing methods implicitly assume that visual tokens maintain consistent semantics across layers and thus indiscriminately inject shallow-layer visual information into deeper layers. We identify a new phenomenon, termed correlation flip, where a visual token positively correlated with the decoding context in an intermediate layer becomes negatively correlated in the final layer, indicating that correct visual evidence is suppressed before prediction. Based on this finding, we propose Correlation-Flip Visual Injection (CFVI), a simple and training-free decoding-time method that selectively identifies correlation-flip tokens and injects corrective visual signals into the final-layer hidden state to restore suppressed evidence and mitigate hallucinations. CFVI can be directly applied to existing LVLMs, and introduces minimal computational overhead. Experiments on multiple benchmarks demonstrate that CFVI achieves strong performance while preserving language fluency.</abstract>
<identifier type="citekey">cao-etal-2026-correlation</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1161/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>26585</start>
<end>26596</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Correlation-Flip Visual Injection for Hallucination Mitigation in Large Vision-Language Models
%A Cao, Jiaxuan
%A Shi, Zeyu
%A Li, Longhua
%A Wang, Lei
%A Qi, Lei
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F cao-etal-2026-correlation
%X Large vision-language models (LVLMs) often hallucinate due to over-reliance on language priors. Existing methods implicitly assume that visual tokens maintain consistent semantics across layers and thus indiscriminately inject shallow-layer visual information into deeper layers. We identify a new phenomenon, termed correlation flip, where a visual token positively correlated with the decoding context in an intermediate layer becomes negatively correlated in the final layer, indicating that correct visual evidence is suppressed before prediction. Based on this finding, we propose Correlation-Flip Visual Injection (CFVI), a simple and training-free decoding-time method that selectively identifies correlation-flip tokens and injects corrective visual signals into the final-layer hidden state to restore suppressed evidence and mitigate hallucinations. CFVI can be directly applied to existing LVLMs, and introduces minimal computational overhead. Experiments on multiple benchmarks demonstrate that CFVI achieves strong performance while preserving language fluency.
%U https://aclanthology.org/2026.emnlp-main.1161/
%P 26585-26596
Markdown (Informal)
[Correlation-Flip Visual Injection for Hallucination Mitigation in Large Vision-Language Models](https://aclanthology.org/2026.emnlp-main.1161/) (Cao et al., EMNLP 2026)
ACL