Low-Rank HOCA: Efficient High-Order Cross-Modal Attention for Video Captioning
This addresses the problem of generating accurate video descriptions by improving cross-modal interactions, though it is incremental as it builds on existing attention-based encoder-decoder structures.
The paper tackles video captioning by proposing Low-Rank HOCA, a method that uses high-order cross-modal attention to capture interactions between multiple modalities and reduces computational costs via tensor decomposition, achieving new state-of-the-art results on MSVD and MSR-VTT datasets.
This paper addresses the challenging task of video captioning which aims to generate descriptions for video data. Recently, the attention-based encoder-decoder structures have been widely used in video captioning. In existing literature, the attention weights are often built from the information of an individual modality, while, the association relationships between multiple modalities are neglected. Motivated by this observation, we propose a video captioning model with High-Order Cross-Modal Attention (HOCA) where the attention weights are calculated based on the high-order correlation tensor to capture the frame-level cross-modal interaction of different modalities sufficiently. Furthermore, we novelly introduce Low-Rank HOCA which adopts tensor decomposition to reduce the extremely large space requirement of HOCA, leading to a practical and efficient implementation in real-world applications. Experimental results on two benchmark datasets, MSVD and MSR-VTT, show that Low-rank HOCA establishes a new state-of-the-art.