CICA：耦合置信度感知预训练与置信度引导注意力，实现鲁棒的多模态情感分析
CICA: Coupling Confidence-Aware Pretraining with Confidence-Informed Attention for Robust Multimodal Sentiment Analysis
把「每个模态编码器自估置信度」作为预训练目标，再用置信度加权注意力做融合，这套机制可直接迁移到视觉-语言任务中处理图像/文本模态噪声或缺失的场景（如噪声图文检索、缺失模态的VQA）。
原文依据 each modality encoder learns to estimate the reliability of its own representation, producing both embeddings and confidence scores
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论多模态情感分析需要整合语言、视觉与声学线索，但这些模态往往带有噪声、不完整或相互矛盾，导致融合不可靠。现有大多数方法假设各模态的可信度一致，因而在信号冲突时性能下降。为此，我们提出 CICA，一个将置信度感知预训练与置信度引导注意力相耦合的框架。在预训练阶段，每个模态编码器学习估计自身表示的可信度，同时输出嵌入与置信度分数。这些分数随后引导一种置信度引导注意力机制，增强可靠模态的贡献、抑制噪声或冲突模态，从而在不同信号条件下实现自适应融合。CICA 在 MOSI、MOSEI、CH-SIMS 与 CH-SIMSv2 四个主要基准上取得最优性能：在 MOSI 上达到 MAE 0.630 与 Corr 0.855，在 MOSEI 上达到 MAE 0.489 与 Corr 0.856，显著超越此前方法；在 Acc-7、Acc-2 与 F1 指标上也观察到一致的提升。在噪声与模态缺失条件下，CICA 的性能保持显著更稳定，表明其鲁棒性与可解释性得到改善。
英文原文摘要 1251 字符
Multimodal sentiment analysis requires integrating language, visual, and acoustic cues, yet these modalities are often noisy, incomplete, or contradictory, making fusion unreliable. Most existing methods assume uniformly trustworthy modalities and thus degrade when signals conflict. To address this, we propose CICA, a framework that couples Confidence-Aware Pretraining with Confidence-Informed Attention. In pretraining, each modality encoder learns to estimate the reliability of its own representation, producing both embeddings and confidence scores. These scores then guide a confidence-informed attention mechanism, which strengthens contributions from reliable modalities while suppressing noisy or conflicting ones, enabling adaptive fusion under varying signal conditions. CICA achieves state-of-the-art performance across four major benchmarks on MOSI, MOSEI, CH-SIMS, and CH-SIMSv2. It achieves MAE 0.630 and Corr 0.855 on MOSI, and MAE 0.489 and Corr 0.856 on MOSEI, significantly surpassing prior methods. Consistent improvements are also observed across Acc-7, Acc-2, and F1 metrics. Under noisy and missing-modality conditions, CICA maintains significantly more stable performance, indicating improved robustness and interpretability.