THOR: Threshold-Based Ranking Loss for Ordinal Regression
This work addresses the problem of improving ordinal classification accuracy for applications requiring precise category predictions, representing an incremental advancement in loss function design.
The paper tackles ordinal regression by introducing a threshold-based pairwise loss function that minimizes regression error to reduce Mean Absolute Error (MAE), achieving the best MAE results on five real-world benchmarks compared to state-of-the-art methods.
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.