THOR: Threshold-Based Ranking Loss for Ordinal Regression

TL;DR

This paper introduces THOR, a novel threshold-based ranking loss for ordinal regression that improves mean absolute error by learning to project inputs onto predefined decision boundaries, validated on real-world benchmarks.

Contribution

The paper proposes a new threshold-based pairwise loss function for ordinal regression, enhancing MAE performance over existing methods.

Findings

Achieves state-of-the-art MAE on five benchmarks.

Introduces a threshold-based loss function for ordinal regression.

Demonstrates effectiveness with CNN feature extraction.

Abstract

In this work, we present a regression-based ordinal regression algorithm for supervised classification of instances into ordinal categories. In contrast to previous methods, in this work the decision boundaries between categories are predefined, and the algorithm learns to project the input examples onto their appropriate scores according to these predefined boundaries. This is achieved by adding a novel threshold-based pairwise loss function that aims at minimizing the regression error, which in turn minimizes the Mean Absolute Error (MAE) measure. We implemented our proposed architecture-agnostic method using the CNN-framework for feature extraction. Experimental results on five real-world benchmarks demonstrate that the proposed algorithm achieves the best MAE results compared to state-of-the-art ordinal regression algorithms.