摘要
为了无缝地适应非平稳数据流分类任务中的不同概念漂移,提出一种基于复制动力学和粒子群优化(Replicator Dynamics and Particle Swarm Optimization,RD-PSO)的自适应数据流分类技术。该技术基于三层体系结构通过从目标数据流的特征池中随机选择一定百分比的特征来创建不同大小的分类类型,使用粒子群优化技术通过单独优化所提算法的每一层中的特征组合来处理突发式和复现式概念漂移。结果表明与现有方法相比,该算法在准确性和鲁棒性方面均优于现有方法。
In order to seamlessly adapt to the different concept drift in the task of non-stationary data flow classification,an adaptive data flow classification technique based on replication dynamics and particle swarm optimization(RD-PSO)is proposed.Based on the three-layer architecture,the classification types of different sizes were created by randomly selecting a certain percentage of features from the feature pool of the target data stream.The improved particle swarm optimization was used to deal with the burst and recurrence concept drift by optimizing the feature combination in each layer of the proposed algorithm separately.The results show that compared with the existing methods,this algorithm is superior to the existing methods in accuracy and robustness.
作者
邹劲松
李芳
Zou Jinsong;Li Fang(School of Putian Big Data Industry,Chongqing Water Resources and Electric Engineering College,Chongqing 402160,China;College of Computer Science,Chongqing University,Chongqing 400044,China)
出处
《计算机应用与软件》
北大核心
2021年第7期246-250,288,共6页
Computer Applications and Software
基金
重庆市教育科学“十三五”规划2017年度重点无经费课题(2017-GX-181)。
关键词
概念漂移
复制动力学
粒子群优化算法
自适应数据流
Concept drifts
Replicator dynamics
Particle swarm optimization algorithm
Adaptive data flow
