Positive semidefinite support vector regression metric learning

TL;DR

This paper introduces two methods to enable positive semidefinite support vector regression in relation alignment metric learning, improving its applicability to complex real-world tasks like multi-label and label distribution learning.

Contribution

It proposes novel methods to incorporate positive semidefinite constraints into SVR-based metric learning, addressing a key limitation of the RAML framework.

Findings

Methods outperform RAML in various classification tasks

Achieve better metric learning performance in multi-label scenarios

Demonstrate effectiveness across different learning paradigms

Abstract

Most existing metric learning methods focus on learning a similarity or distance measure relying on similar and dissimilar relations between sample pairs. However, pairs of samples cannot be simply identified as similar or dissimilar in many real-world applications, e.g., multi-label learning, label distribution learning. To this end, relation alignment metric learning (RAML) framework is proposed to handle the metric learning problem in those scenarios. But RAML framework uses SVR solvers for optimization. It can't learn positive semidefinite distance metric which is necessary in metric learning. In this paper, we propose two methds to overcame the weakness. Further, We carry out several experiments on the single-label classification, multi-label classification, label distribution learning to demonstrate the new methods achieves favorable performance against RAML framework.