Co-learning binary classifiers for LP-based multi-label classification |
| |
Affiliation: | 1. School of Management Science, Qufu Normal University, Rizhao Shandong 276825, China;2. College of Science, China Agricultural University, Beijing 100083, China;3. School of Statistics, Capital University of Economics and Business, Beijing 100070, China |
| |
Abstract: | A simple yet practical multi-label learning method, called label powerset (LP), considers each different combination of labels that appear in the training set as a different class value of a single-label classification task. However, because those classes source from multiple labels, there may be some inherent relationships among them. To tackle this challenge, we propose a novel model which aims to co-learn binary classifiers, by combining the training of binary classifiers and the characterizing the relationship among them into a unified objective function. In addition, we develop an alternating optimization algorithm to solve the proposed problem. Extensive experimental results on various kinds of datasets well demonstrate the effectiveness of the proposed model. |
| |
Keywords: | Multi-label classification Label powerset Binary classifiers Co-learning |
本文献已被 ScienceDirect 等数据库收录! |
|