Единая мультимодальная архитектура кодировщика-декодировщика
method
ID: unifiedmultimodalencoderdecoderarchitecture
Related Papers
Vision-Language Pre-Training for Multimodal Aspect-Based Sentiment Analysis
AbstractAs an important task in sentiment analysis, Multimodal Aspect-Based Sentiment Analysis (MABSA) has attracted increasing attention inrecent years. However, previous approaches either (i) use separately pre-trained visual and textual models, …...