CORE: Learning consistent ordinal representations with convex optimization for image ordinal estimation

Yiming Lei, Zilong Li, Yangyang Li, Junping Zhang, Hongming Shan

Abstract: Highlights•A novel framework for ordinal regression maintaining ordinal manifold in latent space.•The proposed totally ordered set distribution describes ordinal relations resided in ordinal labels.•Ordinal prototype-constrained convex programming and the optimal solutions are derived.•CORE consistently improves performance and interpretability of existing ordinal regression.