About this document

Efficient Low-Rank Multimodal Fusion by TT is a document available to read on EtoBox.

This paper proposes a new method called Low-rank Multimodal Fusion to improve the efficiency of multimodal fusion. Previous tensor-based fusion methods suffer from exponentially increasing computational complexity as more modalities are added. The proposed method uses low-rank tensors to integrate multiple modalities into a compact representation, reducing parameters and runtime. It is evaluated on three tasks - sentiment analysis, speaker trait analysis, and emotion recognition - achieving competitive resu

Author

TT

Language

EN