HMEPR: A Hierarchical Multimodal Emotion Recognition Framework With Redundancy-Aware Feature Refinement
Multimodal emotion recognition (MER) aims to infer affective states from heterogeneous inputs such as text, audio, and visual signals. In practice, however, redundant and conflicting information across modalities can weaken cross-modal alignment and impair downstream fusion, especially when emotiona...
Сохранить в:
| Главные авторы: | , , , , |
|---|---|
| Формат: | Artigo |
| Язык: | Inglês |
| Опубликовано: |
IEEE
2026-01-01
|
| Серии: | IEEE Access |
| Предметы: | |
| Online-ссылка: | https://ieeexplore.ieee.org/document/11515149/ |
| Метки: |
Нет меток, Требуется 1-ая метка записи!
|