MMoLRE, a low-rank mixture-of-experts model for joint multimodal sentiment and emotion prediction, reports state-of-the-art sentiment scores on CMU-MOSI and CMU-MOSEI while emotion recognition remains competitive but not best.
Hybrid contrastive learning of tri-modal representation for multimodal sentiment analysis,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.AI 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Multimodal Mixture of Low-Rank Experts for Sentiment Analysis and Emotion Recognition
MMoLRE, a low-rank mixture-of-experts model for joint multimodal sentiment and emotion prediction, reports state-of-the-art sentiment scores on CMU-MOSI and CMU-MOSEI while emotion recognition remains competitive but not best.