@inproceedings{peng-etal-2026-dual,
title = "Dual-Group Exploration: Reinforcement Learning for Video Understanding in Multimodal Large Language Models",
author = "Peng, Qucheng and
Hou, Rui and
Zhang, Jingyi and
Zhu, Shengjie and
Xu, Yingfu and
Xu, Haotian and
Vesdapunt, Noranart and
Chen, Chen and
Zhou, Ning",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.835/",
doi = "10.18653/2026.findings-emnlp.835",
pages = "17696--17719",
ISBN = "979-8-89176-462-0",
abstract = "Reinforcement Learning (RL) has significantly enhanced reasoning in Large Language Models. However, applying state-of-the-art RL methods like Group Relative Policy Optimization (GRPO) to video-based Multimodal LLMs (MLLMs) remains sub-optimal. We identify two key exploration bottlenecks: (1) Advantage Vanishing: generating multiple responses from invariant video contexts leads to near-uniform rewards, neutralizing learning signals after in-group standardization; (2) Visual Overfitting: standard RL lacks mechanisms to regularize sensitivity to visual perturbations under single-epoch training, causing policies to overfit to sparse temporal cues. To address these, we propose Dual-Group Exploration, a framework revitalizing learning signals at both input and output levels. At the input level, a Dual-Group Sampling strategy breaks visual determinism via complementary temporal sequences. At the output level, an Entropy-rated ELO Advantage captures inter-group quality variances. Concurrently, a Cross-Visual Counterfactual Weight establishes a Visual Consistency Region, safely driving exploration by enforcing reasoning consistency across varied views. Achieving state-of-the-art performance across six video reasoning benchmarks, our method demonstrates that consistency-guided exploration is beneficial for video understanding."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="peng-etal-2026-dual">
<titleInfo>
<title>Dual-Group Exploration: Reinforcement Learning for Video Understanding in Multimodal Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Qucheng</namePart>
<namePart type="family">Peng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rui</namePart>
<namePart type="family">Hou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jingyi</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shengjie</namePart>
<namePart type="family">Zhu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yingfu</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haotian</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Noranart</namePart>
<namePart type="family">Vesdapunt</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chen</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ning</namePart>
<namePart type="family">Zhou</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>Findings of the Association for Computational Linguistics: EMNLP 2026</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">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Reinforcement Learning (RL) has significantly enhanced reasoning in Large Language Models. However, applying state-of-the-art RL methods like Group Relative Policy Optimization (GRPO) to video-based Multimodal LLMs (MLLMs) remains sub-optimal. We identify two key exploration bottlenecks: (1) Advantage Vanishing: generating multiple responses from invariant video contexts leads to near-uniform rewards, neutralizing learning signals after in-group standardization; (2) Visual Overfitting: standard RL lacks mechanisms to regularize sensitivity to visual perturbations under single-epoch training, causing policies to overfit to sparse temporal cues. To address these, we propose Dual-Group Exploration, a framework revitalizing learning signals at both input and output levels. At the input level, a Dual-Group Sampling strategy breaks visual determinism via complementary temporal sequences. At the output level, an Entropy-rated ELO Advantage captures inter-group quality variances. Concurrently, a Cross-Visual Counterfactual Weight establishes a Visual Consistency Region, safely driving exploration by enforcing reasoning consistency across varied views. Achieving state-of-the-art performance across six video reasoning benchmarks, our method demonstrates that consistency-guided exploration is beneficial for video understanding.</abstract>
<identifier type="citekey">peng-etal-2026-dual</identifier>
<identifier type="doi">10.18653/2026.findings-emnlp.835</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.835/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>17696</start>
<end>17719</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Dual-Group Exploration: Reinforcement Learning for Video Understanding in Multimodal Large Language Models
%A Peng, Qucheng
%A Hou, Rui
%A Zhang, Jingyi
%A Zhu, Shengjie
%A Xu, Yingfu
%A Xu, Haotian
%A Vesdapunt, Noranart
%A Chen, Chen
%A Zhou, Ning
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F peng-etal-2026-dual
%X Reinforcement Learning (RL) has significantly enhanced reasoning in Large Language Models. However, applying state-of-the-art RL methods like Group Relative Policy Optimization (GRPO) to video-based Multimodal LLMs (MLLMs) remains sub-optimal. We identify two key exploration bottlenecks: (1) Advantage Vanishing: generating multiple responses from invariant video contexts leads to near-uniform rewards, neutralizing learning signals after in-group standardization; (2) Visual Overfitting: standard RL lacks mechanisms to regularize sensitivity to visual perturbations under single-epoch training, causing policies to overfit to sparse temporal cues. To address these, we propose Dual-Group Exploration, a framework revitalizing learning signals at both input and output levels. At the input level, a Dual-Group Sampling strategy breaks visual determinism via complementary temporal sequences. At the output level, an Entropy-rated ELO Advantage captures inter-group quality variances. Concurrently, a Cross-Visual Counterfactual Weight establishes a Visual Consistency Region, safely driving exploration by enforcing reasoning consistency across varied views. Achieving state-of-the-art performance across six video reasoning benchmarks, our method demonstrates that consistency-guided exploration is beneficial for video understanding.
%R 10.18653/2026.findings-emnlp.835
%U https://aclanthology.org/2026.findings-emnlp.835/
%U https://doi.org/10.18653/2026.findings-emnlp.835
%P 17696-17719
Markdown (Informal)
[Dual-Group Exploration: Reinforcement Learning for Video Understanding in Multimodal Large Language Models](https://aclanthology.org/2026.findings-emnlp.835/) (Peng et al., Findings 2026)
ACL
- Qucheng Peng, Rui Hou, Jingyi Zhang, Shengjie Zhu, Yingfu Xu, Haotian Xu, Noranart Vesdapunt, Chen Chen, and Ning Zhou. 2026. Dual-Group Exploration: Reinforcement Learning for Video Understanding in Multimodal Large Language Models. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 17696–17719, Budapest, Hungary. Association for Computational Linguistics.