基于相似度的多标签分类算法优化  

Optimization of Multi-label Classification Algorithm Based on Similarity

作　　者：刘云[1] 肖添 肖雪 LIU Yun;XIAO Tian;XIAO Xue(School of Information Engineering and Automation,Kunming University of Science and Technology,Kunming 650500)

机构地区：[1]昆明理工大学信息工程与自动化学院,昆明650500

出　　处：《计算机与数字工程》2022年第2期243-246,260,共5页Computer & Digital Engineering

基　　金：国家自然科学基金项目(编号:61262040)资助。

摘　　要：多标签学习是分类任务中一个重要研究方向,如何保证高分类精度是关键要素。论文提出基于相似度的多标签分类算法SMLC。该算法首先构建实例相似度函数,再采用并行计算方式算出相似值,最后通过加权计算类标签集合权重或者学习阈值方法预测类标签集合。仿真结果表明,与RankSVM、ML-KNN算法对比,SMLC在多标签分类任务的多个指标上表现更优。Multi-label classification is an important learning problem in many fields,and high accuracy is a key factor in classification. The paper proposes a multi-label classification algorithm SMLC based on similarity. The algorithm builds a similarity function for instances,and calculates similarity values in parallel at the same time. It uses decision function weighted calculation or predicts a set of instance class labels by learning a threshold function. Simulation results show that,compared with RankSVM and ML-KNN algorithms,SMLC is optimized on multiple indicators for multi-label classification.

关 键 词：多标签分类 精度 相似度 SMLC 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]