Granular Multimodal Attention Networks for Visual Dialog
This work addresses a specific bottleneck in multimodal AI for visual dialog, offering incremental improvements in attention mechanisms.
The authors tackled the problem of determining the optimal granularity for attention mechanisms in visual dialog tasks, proposing a Granular Multi-modal Attention method that improved both image and text attention networks and achieved the best performance by jointly attending to image and text granules.
Vision and language tasks have benefited from attention. There have been a number of different attention models proposed. However, the scale at which attention needs to be applied has not been well examined. Particularly, in this work, we propose a new method Granular Multi-modal Attention, where we aim to particularly address the question of the right granularity at which one needs to attend while solving the Visual Dialog task. The proposed method shows improvement in both image and text attention networks. We then propose a granular Multi-modal Attention network that jointly attends on the image and text granules and shows the best performance. With this work, we observe that obtaining granular attention and doing exhaustive Multi-modal Attention appears to be the best way to attend while solving visual dialog.