Deep Ordinal Regression with Label Diversity

TL;DR

This paper introduces a novel deep ordinal regression method that uses multiple discrete data representations simultaneously, improving prediction accuracy in regression tasks by leveraging label diversity.

Contribution

It proposes an end-to-end differentiable approach that extends traditional regression via classification by incorporating multiple label discretizations.

Findings

Reduces prediction error compared to baseline RvC methods

Maintains similar model complexity while improving accuracy

Effective on three challenging regression tasks

Abstract

Regression via classification (RvC) is a common method used for regression problems in deep learning, where the target variable belongs to a set of continuous values. By discretizing the target into a set of non-overlapping classes, it has been shown that training a classifier can improve neural network accuracy compared to using a standard regression approach. However, it is not clear how the set of discrete classes should be chosen and how it affects the overall solution. In this work, we propose that using several discrete data representations simultaneously can improve neural network learning compared to a single representation. Our approach is end-to-end differentiable and can be added as a simple extension to conventional learning methods, such as deep neural networks. We test our method on three challenging tasks and show that our method reduces the prediction error compared to a baseline RvC approach while maintaining a similar model complexity.