Target-Oriented Deformation of Visual-Semantic Embedding Space
arXiv:1910.06514
Abstract
Multimodal embedding is a crucial research topic for cross-modal understanding, data mining, and translation. Many studies have attempted to extract representations from given entities and align them in a shared embedding space. However, because entities in different modalities exhibit different abstraction levels and modality-specific information, it is insufficient to embed related entities close to each other. In this study, we propose the Target-Oriented Deformation Network (TOD-Net), a novel module that continuously deforms the embedding space into a new space under a given condition, thereby adjusting similarities between entities. Unlike methods based on cross-modal attention, TOD-Net is a post-process applied to the embedding space learned by existing embedding systems and improves their performances of retrieval. In particular, when combined with cutting-edge models, TOD-Net gains the state-of-the-art cross-modal retrieval model associated with the MSCOCO dataset. Qualitative analysis reveals that TOD-Net successfully emphasizes entity-specific concepts and retrieves diverse targets via handling higher levels of diversity than existing models.
arXiv comment8 pages
References in corpus (9)
- Efficient Estimation of Word Representations in Vector Space
- Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
- Unifying Visual-Semantic Embeddings with Multimodal Neural Language Models
- Skip-Thought Vectors
- Poincaré Embeddings for Learning Hierarchical Representations
- Finding beans in burgers: Deep semantic-visual embedding with localization
- Learning Visual Relation Priors for Image-Text Matching and Image Captioning with Neural Scene Graph Generators
- Gaussian Word Embedding with a Wasserstein Distance Loss
- Instance-aware Image and Sentence Matching with Selective Multimodal LSTM