Alert button

Sequential Late Fusion Technique for Multi-modal Sentiment Analysis

Jun 22, 2021
Debapriya Banerjee, Fotios Lygerakis, Fillia Makedon

Figure 1 for Sequential Late Fusion Technique for Multi-modal Sentiment Analysis
Figure 2 for Sequential Late Fusion Technique for Multi-modal Sentiment Analysis

Share this with someone who'll enjoy it:

Multi-modal sentiment analysis plays an important role for providing better interactive experiences to users. Each modality in multi-modal data can provide different viewpoints or reveal unique aspects of a user's emotional state. In this work, we use text, audio and visual modalities from MOSI dataset and we propose a novel fusion technique using a multi-head attention LSTM network. Finally, we perform a classification task and evaluate its performance.

* 2 pages, 1 figure, 1 table  
View paper onarxiv icon

Share this with someone who'll enjoy it: