The More You Know: Using Knowledge Graphs for Image Classification

TL;DR

This paper explores integrating knowledge graphs into image classification to enhance performance, inspired by human reasoning, by introducing a novel Graph Search Neural Network that efficiently incorporates large structured knowledge bases.

Contribution

It introduces the Graph Search Neural Network, a new method for embedding large knowledge graphs into vision models for improved classification accuracy.

Findings

Outperforms standard neural network baselines in multi-label classification

Effectively incorporates large knowledge graphs into vision pipelines

Demonstrates the benefit of structured prior knowledge in image understanding

Abstract

One characteristic that sets humans apart from modern learning-based computer vision algorithms is the ability to acquire knowledge about the world and use that knowledge to reason about the visual world. Humans can learn about the characteristics of objects and the relationships that occur between them to learn a large variety of visual concepts, often with few examples. This paper investigates the use of structured prior knowledge in the form of knowledge graphs and shows that using this knowledge improves performance on image classification. We build on recent work on end-to-end learning on graphs, introducing the Graph Search Neural Network as a way of efficiently incorporating large knowledge graphs into a vision classification pipeline. We show in a number of experiments that our method outperforms standard neural network baselines for multi-label classification.