Papers with Supervised Angular-based Contrastive Learning
Improving Multimodal Sentiment Analysis: Supervised Angular margin-based Contrastive Learning for Enhanced Fusion Representation (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for multimodal sentiment analysis focus on forming positive and negative pairs, neglecting the variation in sentiment scores within the same class. |
| Approach: | They propose a framework to enhance discrimination and generalizability of the multimodal representation and overcome biases in the fusion vector’s modality. |
| Outcome: | The proposed model improves discrimination and generalizability of the multimodal representation and overcomes biases in the fusion vector’s modality. |