Order-Embeddings of Images and Language

TL;DR

This paper introduces a method for learning ordered representations that model the hierarchical relationships among images and language, improving performance in tasks like hypernym prediction and image-caption retrieval.

Contribution

It proposes a general approach for explicitly modeling the partial order structure in visual-semantic hierarchies, applicable across multiple tasks.

Findings

Improved hypernym prediction accuracy

Enhanced image-caption retrieval performance

Effective modeling of visual-semantic hierarchies

Abstract

Hypernymy, textual entailment, and image captioning can be seen as special cases of a single visual-semantic hierarchy over words, sentences, and images. In this paper we advocate for explicitly modeling the partial order structure of this hierarchy. Towards this goal, we introduce a general method for learning ordered representations, and show how it can be applied to a variety of tasks involving images and language. We show that the resulting representations improve performance over current approaches for hypernym prediction and image-caption retrieval.