Controlling Class Layout for Deep Ordinal Classification via Constrained Proxies Learning

TL;DR

This paper introduces a novel Constrained Proxies Learning (CPL) method for deep ordinal classification, which learns and constrains class proxies to achieve an ideal ordinal layout in feature space, improving classification performance.

Contribution

The paper proposes a new CPL approach with hard and soft layout constraints to enforce ordinal class arrangements in feature space, enhancing deep ordinal classification.

Findings

CPL outperforms previous methods in experiments.

Hard and soft constraints effectively enforce ordinal layouts.

Improved class separation and ordering in feature space.

Abstract

For deep ordinal classification, learning a well-structured feature space specific to ordinal classification is helpful to properly capture the ordinal nature among classes. Intuitively, when Euclidean distance metric is used, an ideal ordinal layout in feature space would be that the sample clusters are arranged in class order along a straight line in space. However, enforcing samples to conform to a specific layout in the feature space is a challenging problem. To address this problem, in this paper, we propose a novel Constrained Proxies Learning (CPL) method, which can learn a proxy for each ordinal class and then adjusts the global layout of classes by constraining these proxies. Specifically, we propose two kinds of strategies: hard layout constraint and soft layout constraint. The hard layout constraint is realized by directly controlling the generation of proxies to force them to be placed in a strict linear layout or semicircular layout (i.e., two instantiations of strict ordinal layout). The soft layout constraint is realized by constraining that the proxy layout should always produce unimodal proxy-to-proxies similarity distribution for each proxy (i.e., to be a relaxed ordinal layout). Experiments show that the proposed CPL method outperforms previous deep ordinal classification methods under the same setting of feature extractor.