MUTAN: Multimodal Tucker Fusion for Visual Question Answering
This work addresses efficiency and interpretability challenges in multimodal fusion for VQA, offering an incremental improvement over existing architectures.
The paper tackled the high dimensionality issue in bilinear models for Visual Question Answering by introducing MUTAN, a multimodal Tucker decomposition method, which achieved state-of-the-art results on VQA tasks.
Bilinear models provide an appealing framework for mixing and merging information in Visual Question Answering (VQA) tasks. They help to learn high level associations between question meaning and visual concepts in the image, but they suffer from huge dimensionality issues. We introduce MUTAN, a multimodal tensor-based Tucker decomposition to efficiently parametrize bilinear interactions between visual and textual representations. Additionally to the Tucker framework, we design a low-rank matrix-based decomposition to explicitly constrain the interaction rank. With MUTAN, we control the complexity of the merging scheme while keeping nice interpretable fusion relations. We show how our MUTAN model generalizes some of the latest VQA architectures, providing state-of-the-art results.