InterMulti:Multi-view Multimodal Interactions with Text-dominated Hierarchical High-order Fusion for Emotion Analysis
원문 초록
Humans are sophisticated at reading interlocutors' emotions from multimodal signals, such as speech contents, voice tones and facial expressions. However, machines might struggle to understand various emotions due to the difficulty of effectively decoding emotions from the complex interactions between multimodal signals. In this paper, we propose a multimodal emotion analysis framework, InterMulti, to capture complex multimodal interactions from different views and identify emotions from multimodal signals. Our proposed framework decomposes signals of different modalities into three kinds of multimodal interaction representations, including a modality-full interaction representation, a modality-shared interaction representation, and three modality-specific interaction representations. Additionally, to balance the contribution of different modalities and learn a more informative latent interaction representation, we developed a novel Text-dominated Hierarchical High-order Fusion(THHF) module. THHF module reasonably integrates the above three kinds of representations into a comprehensive multimodal interaction representation. Extensive experimental results on widely used datasets, (i.e.) MOSEI, MOSI and IEMOCAP, demonstrate that our method outperforms the state-of-the-art.
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
Long-Short Distance Graph Neural Networks and Improved Curriculum Learning for Emotion Recognition in Conversation
Dynamic Parameter Memory: Temporary LoRA-Enhanced LLM for Long-Sequence Emotion Recognition in Conversation
Exploring Remote Physiological Signal Measurement under Dynamic Lighting Conditions at Night: Dataset, Experiment, and Analysis
Infant Cry Emotion Recognition Using Improved ECAPA-TDNN with Multiscale Feature Fusion and Attention Enhancement
Stream-Omni: Simultaneous Multimodal Interactions with Large Language-Vision-Speech Model
Towards Robust Multimodal Emotion Recognition under Missing Modalities and Distribution Shifts
토론
이 논문과 연결된 포럼 글이 아직 없습니다.