Does Seeing More Mean Knowing More? Mono-Anchored Advantage Normalization for Multi-Source Visual Reasoning

• URL: http://arxiv.org/abs/2605.25437v1
• Date: Mon, 25 May 2026 05:29:07 GMT
• Title: Does Seeing More Mean Knowing More? Mono-Anchored Advantage Normalization for Multi-Source Visual Reasoning
• Abstract summary: We propose a novel mono-anchored multi-source reasoning framework that models each visual modality as an independent information source.<n>Our method effectively quantifies information gain introduced by multi-source integration in gradient estimation.<n> Empirical results also show impressive 3.2% and 4.9% performance gains on GRPO and DAPO.
• Score: 36.976702924164584
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Visual reasoning through reinforcement learning with verifiable rewards (RLVR) has achieved remarkable progress. However, when dealing with multi-source inputs, existing approaches tend to treat them as a mere accumulation of information, lacking explicit mechanisms to distinguish whether integrating additional sources yields information gain or introduces interference. Therefore, they struggle to effectively model dynamic interaction when integrating multiple sources, particularly when they differ significantly in physical properties and semantics, e.g., infrared and depth, leading to inferior performance to mono-source reasoning when a certain source holds the dominant signal. To address this issue, we propose MARS, a novel mono-anchored multi-source reasoning framework that models each visual modality as an independent information source. Specifically, by treating mono-source rewards as dynamic anchors, our method explicitly incorporates the information gain introduced by multi-source fusion into advantage normalization and adaptively emphasizes mutual promotion between sources while suppressing potential noise or conflicts during RLVR. From theoretical analysis, our method effectively quantifies information gain introduced by multi-source integration in gradient estimation, enabling consistent modality regulation. Empirical results also show impressive 3.2% and 4.9% performance gains on GRPO and DAPO across diverse datasets, confirming effectiveness of our method.

Related papers

• MultAttnAttrib: Training-Free Multimodal Attribution in Long Document Question Answering [73.47440561592556]
  We introduce MultAttnAttrib, a training-free attribution-generation method.<n>We show that MultAttnAttrib consistently outperforms a variety of attribution-generation methods.<n>Our method produces attributions at up to one-seventh of the direct inference latency compared to prompting.
  arXiv  Detail & Related papers  (2026-07-01T19:29:19Z)
• Learning Invariant Modality Representation for Robust Multimodal Learning from a Causal Inference Perspective [6.3310165899037045]
  We propose a causal modality-invariant representation (CmIR) learning framework for robust multimodal learning.<n>CmIR ensures that the learned invariant representations retain stable predictive relationships with labels across different environments.<n>CmIR excels on out-of-distribution data and noisy data, confirming its robustness and generalizability.
  arXiv  Detail & Related papers  (2026-04-20T16:16:36Z)
• Orthogonalized Multimodal Contrastive Learning with Asymmetric Masking for Structured Representations [4.67724003380452]
  Multimodal learning seeks to integrate information from heterogeneous sources, where signals may be shared across modalities, specific to individual modalities, or emerge only through their interaction.<n>While self-supervised multimodal contrastive learning has achieved remarkable progress, most existing methods predominantly capture redundant cross-modal signals, often neglecting modality-specific (unique) and interaction-driven (synergistic) information.<n>Recent extensions broaden this perspective, yet they either fail to explicitly model synergistic interactions or learn different information components in an entangled manner, leading to incomplete representations and potential information leakage.<n>We introduce textbfCOrAL, a principled framework
  arXiv  Detail & Related papers  (2026-02-16T18:06:53Z)
• MMR1: Enhancing Multimodal Reasoning with Variance-Aware Sampling and Open Resources [113.33902847941941]
  Variance-Aware Sampling (VAS) is a data selection strategy guided by Variance Promotion Score (VPS)<n>We release large-scale, carefully curated resources containing 1.6M long CoT cold-start data and 15k RL QA pairs.<n> Experiments across mathematical reasoning benchmarks demonstrate the effectiveness of both the curated data and the proposed VAS.
  arXiv  Detail & Related papers  (2025-09-25T14:58:29Z)
• Disentangling Bias by Modeling Intra- and Inter-modal Causal Attention for Multimodal Sentiment Analysis [25.791796193062012]
  Multimodal sentiment analysis (MSA) aims to understand human emotions by integrating information from multiple modalities, such as text, audio, and visual data.<n>Existing methods often suffer from spurious correlations both within and across modalities, leading models to rely on statistical shortcuts rather than true causal relationships.<n>We propose a Multi-relational Multimodal Causal Intervention (MMCI) model, which leverages the backdoor adjustment from causal theory to address the confounding effects of such shortcuts.
  arXiv  Detail & Related papers  (2025-08-07T03:24:04Z)
• Asymmetric Reinforcing against Multi-modal Representation Bias [59.685072206359855]
  We propose an Asymmetric Reinforcing method against Multimodal representation bias (ARM)<n>Our ARM dynamically reinforces the weak modalities while maintaining the ability to represent dominant modalities through conditional mutual information.<n>We have significantly improved the performance of multimodal learning, making notable progress in mitigating imbalanced multimodal learning.
  arXiv  Detail & Related papers  (2025-01-02T13:00:06Z)
• GAMED: Knowledge Adaptive Multi-Experts Decoupling for Multimodal Fake News Detection [18.157900272828602]
  Multimodal fake news detection often involves modelling heterogeneous data sources, such as vision and language.<n>This paper develops a significantly novel approach, GAMED, for multimodal modelling.<n>It focuses on generating distinctive and discriminative features through modal decoupling to enhance cross-modal synergies.
  arXiv  Detail & Related papers  (2024-12-11T19:12:22Z)
• Interpretable Multi-Source Data Fusion Through Latent Variable Gaussian Process [8.207427766052044]
  The proposed approach is demonstrated on and analyzed through two mathematical and two materials science case studies. It is observed that compared to using single-source and source unaware machine learning models, the proposed multi-source data fusion framework can provide better predictions for sparse-data problems.
  arXiv  Detail & Related papers  (2024-02-06T16:54:59Z)
• Mitigating Shortcut Learning with Diffusion Counterfactuals and Diverse Ensembles [104.60508550106618]
  We propose DiffDiv, an ensemble diversification framework exploiting Diffusion Probabilistic Models (DPMs)<n>We show that DPMs can generate images with novel feature combinations, even when trained on samples displaying correlated input features.<n>We show that DPM-guided diversification is sufficient to remove dependence on shortcut cues, without a need for additional supervised signals.
  arXiv  Detail & Related papers  (2023-11-23T15:47:33Z)
• Generalized Product-of-Experts for Learning Multimodal Representations in Noisy Environments [18.14974353615421]
  We propose a novel method for multimodal representation learning in a noisy environment via the generalized product of experts technique. In the proposed method, we train a separate network for each modality to assess the credibility of information coming from that modality. We attain state-of-the-art performance on two challenging benchmarks: multimodal 3D hand-pose estimation and multimodal surgical video segmentation.
  arXiv  Detail & Related papers  (2022-11-07T14:27:38Z)
• Variational Distillation for Multi-View Learning [104.17551354374821]
  We design several variational information bottlenecks to exploit two key characteristics for multi-view representation learning. Under rigorously theoretical guarantee, our approach enables IB to grasp the intrinsic correlation between observations and semantic labels.
  arXiv  Detail & Related papers  (2022-06-20T03:09:46Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.