Abstract
Multimodal sentiment analysis aims to understand human emotions by integrating textual, auditory, and visual modalities. Although Multimodal Large Language Models (MLLMs) have achieved state-of-the-art performance via supervised fine-tuning (SFT), their end-to-end "black-box" nature limits interpretability. Existing methods incorporating Chain-of-Thought (CoT) reasoning are hindered by high annotation costs, while Reinforcement Learning (RL) faces challenges such as low exploration efficiency and sparse rewards, particularly on hard samples. To address these issues, we propose a novel training framework that integrates structured Discrimination-Calibration (DC) reasoning with Hint-based Reinforcement Learning. First, we perform cold-start SFT using high-quality CoT data synthesized by a teacher model (Qwen3Omni-30B), which inherently contains the DC structure. This equips the model with a reasoning paradigm that performs macro discrimination followed by fine-grained calibration from the initial stage. Building on this, we propose Hint-GRPO, which leverages the discrimination phase within the DC structure as a verifiable anchor during RL to provide directional hints for hard samples, guiding policy optimization and effectively mitigating the reward sparsity problem. Experiments on the Qwen2.5Omni-7B model demonstrate that our method not only achieves higher accuracy in fine-grained sentiment regression tasks but also generates high-quality structured reasoning chains. Crucially, it exhibits superior generalization capability in cross-domain evaluations. This enhances model interpretability while validating the positive contribution of explicit reasoning steps to model robustness, offering a new paradigm for building trustworthy and efficient sentiment analysis systems.
Abstract (translated)
多模态情感分析旨在通过整合文本、听觉和视觉模态来理解人类情感。尽管多模态大语言模型(MLLMs)已通过监督微调(SFT)达到最先进性能,但其端到端的“黑箱”特性限制了可解释性。现有结合思维链(CoT)推理的方法受限于高昂的标注成本,而强化学习(RL)则面临探索效率低、奖励稀疏等挑战,尤其在困难样本上表现明显。针对这些问题,我们提出一种新颖的训练框架,将结构化的判别-校准(DC)推理与基于提示的强化学习相结合。首先,我们使用教师模型(Qwen3Omni-30B)合成的高质量CoT数据进行冷启动SFT,这些数据天然包含DC结构,使模型从初始阶段就具备“宏观判别+细粒度校准”的推理范式。在此基础上,我们提出Hint-GRPO方法,利用DC结构中的判别阶段作为RL中的可验证锚点,为困难样本提供方向性提示,引导策略优化,有效缓解奖励稀疏问题。在Qwen2.5Omni-7B模型上的实验表明,我们的方法不仅在细粒度情感回归任务中取得更高准确率,还能生成高质量的结构化推理链,并在跨域评估中展现出卓越的泛化能力。这既提升了模型可解释性,也验证了显式推理步骤对模型鲁棒性的积极贡献,为构建可信高效的情感分析系统提供了新范式。
URL
https://arxiv.org/abs/2604.00013