Multimodal intent recognition is aimed at understanding user intentions by integrating information from multiple modalities. It has attracted increasing attention in recently developed dialog systems. The existing studies have focused mainly on modeling semantic interactions within and across modalities, but they often overlook the reliability of each modality. In real-world scenarios, inputs may be corrupted by noisy audio, blurred or occluded videos, or ambiguous text, making it difficult for the employed model to determine who to trust and how much to trust. To address this challenge, we propose a method called explicit confidence-focused multimodal intent recognition (ECFMIR). The core idea of this approach is to assign each modality and each cross-modal associations feature a dedicated confidence lens (CLens) that explicitly estimates the confidence level in a hypothetical manner. This design helps reduce the degree of uncertainty and mitigate the risk of incorrect predictions when addressing conflicting inputs. Comprehensive experiments conducted on two benchmark multimodal intent recognition datasets demonstrate the effectiveness of our method. A further analysis reveals that ECFMIR achieves significant advantages for high-conflict categories and under low-resource conditions.

Multimodal intent detection leverages complementary information from diverse sensors to achieve precise semantic understanding; however, existing methodologies predominantly operate under the ideal assumption of modality completeness. In practical deployments, missing modalities, stemming from sensor failure, background noise, or privacy constraints, are inevitable and uncertain, leading to severe performance degradation. To address this challenge, we propose SADF, a Structure Aware Distillation Framework designed to facilitate robust cross-level knowledge transfer from a full-modality teacher to a partial-modality student. At the semantic level, semantic topology distillation aligns prototype similarity distributions between the teacher and the student, capturing class topology and hierarchical relations. At the discriminative level, boundary aware distillation decouples predictions into target and nontarget classes, enforcing alignment with the teacher to enhance discriminability and suppress noise. By integrating these strategies, SADF enables the student to learn robust representations and decisions when uncertain modalities are missing. The results of experiments based on two benchmarks demonstrate that SADF consistently outperforms strong baselines.