Learning Zero-Shot Multifaceted Visually Grounded Word Embeddings via Multi-Task Training
This work addresses the challenge of linking language to perceptual knowledge for natural language processing and computer vision applications, offering a novel approach that improves performance on both abstract and concrete words.
The paper tackles the problem of language grounding by proposing a method to implicitly ground word embeddings via multi-task learning, preserving abstract linguistic knowledge while integrating visual information. The resulting embeddings outperform previous works on multiple benchmarks and show strong correlation with human judgments.
Language grounding aims at linking the symbolic representation of language (e.g., words) into the rich perceptual knowledge of the outside world. The general approach is to embed both textual and visual information into a common space -the grounded space-confined by an explicit relationship between both modalities. We argue that this approach sacrifices the abstract knowledge obtained from linguistic co-occurrence statistics in the process of acquiring perceptual information. The focus of this paper is to solve this issue by implicitly grounding the word embeddings. Rather than learning two mappings into a joint space, our approach integrates modalities by determining a reversible grounded mapping between the textual and the grounded space by means of multi-task learning. Evaluations on intrinsic and extrinsic tasks show that our embeddings are highly beneficial for both abstract and concrete words. They are strongly correlated with human judgments and outperform previous works on a wide range of benchmarks. Our grounded embeddings are publicly available here.