Learning Interclass Relations for Image Classification

TL;DR

This paper introduces methods to incorporate interclass relations into image classification, reducing data requirements and improving performance, especially in medical imaging where labeled data is scarce.

Contribution

It presents novel formulations that embed interclass relations either through manual knowledge or learned representations, enhancing classification efficiency.

Findings

Improved classification accuracy in medical imaging tasks.

Reduced data needs by leveraging interclass relations.

Effective encoding of natural class relationships in label representations.

Abstract

In standard classification, we typically treat class categories as independent of one-another. In many problems, however, we would be neglecting the natural relations that exist between categories, which are often dictated by an underlying biological or physical process. In this work, we propose novel formulations of the classification problem, based on a realization that the assumption of class-independence is a limiting factor that leads to the requirement of more training data. First, we propose manual ways to reduce our data needs by reintroducing knowledge about problem-specific interclass relations into the training process. Second, we propose a general approach to jointly learn categorical label representations that can implicitly encode natural interclass relations, alleviating the need for strong prior assumptions, which are not always available. We demonstrate this in the domain of medical images, where access to large amounts of labelled data is not trivial. Specifically, our experiments show the advantages of this approach in the classification of Intravenous Contrast enhancement phases in CT images, which encapsulate multiple interesting inter-class relations.