Meta-RoMSA: A Reciprocal Translation and Modality Balancing Framework for Robust Multimodal Sentiment Analysis Using Meta-Learning

Multimodal sentiment analysis has demonstrated significant application potential. However, it faces critical challenges, including modal asynchrony, missing modalities, and noise interference. Existing methods exhibit notable limitations: they lack effective mechanisms for handling fine-grained rand...

Volledige beschrijving

Bibliografische gegevens

Hoofdauteurs:	Lei Pan, Hongqiao Li
Formaat:	Artigo
Taal:	Inglês
Gepubliceerd in:	IEEE 2026-01-01
Reeks:	IEEE Access
Onderwerpen:	Multimodal sentiment analysis robust modality translation meta-learning unimodal weight
Online toegang:	https://ieeexplore.ieee.org/document/11321098/
Tags:	Voeg label toe Geen labels, Wees de eerste die dit record labelt!