Similarity-based Multi-label Learning

TL;DR

This paper introduces SML, a similarity-based method for multi-label classification and label set size prediction, demonstrating its effectiveness across various datasets and outperforming many existing algorithms.

Contribution

The paper presents a novel similarity-based approach for multi-label learning and label set size prediction, advancing the state-of-the-art in multi-label classification.

Findings

SML outperforms existing algorithms on multiple evaluation metrics.

The similarity-based approach effectively predicts label set sizes.

Experimental results validate the method's robustness across datasets.

Abstract

Multi-label classification is an important learning problem with many applications. In this work, we propose a principled similarity-based approach for multi-label learning called SML. We also introduce a similarity-based approach for predicting the label set size. The experimental results demonstrate the effectiveness of SML for multi-label classification where it is shown to compare favorably with a wide variety of existing algorithms across a range of evaluation criterion.