基于SVM算法的海洋微生物种群多样性信息自适应分类方法  

Adaptive Classification Method for Marine Microbial Population Diversity Information Based on SVM Algorithm

作　　者：张随艳 ZHANG Suiyan(The Eighth Geological Brigade of Hebei Bureau of Geological and Mineral Exploration and Development(Marine Geological Resource Survey Center of Heber Province),Qinhuangdao 066000,China)

机构地区：[1]河北省地质矿产勘查开发局第八地质大队(河北省海洋地质资源调查中心),河北秦皇岛066000

出　　处：《海洋技术学报》2024年第5期36-42,共7页Journal of Ocean Technology

摘　　要：为了可以精准且高效地实现海洋微生物种群多样性信息分类,本文提出一种基于支持向量机(Support Vector Machine,SVM)算法的海洋微生物种群多样性信息自适应分类方法。通过无线传感器网络展开海洋微生物种群信息采集点的部署,建立海洋微生物种群信息采集信道模型,在带宽受距离约束条件下设计载频频率编码,利用波特间隔均衡控制方法达到海洋微生物种群信息采集优化的目的。在核空间中,对采集的海洋微生物种群信息展开谱聚类,确定具有代表性的信息点,同时引入SVM对全部样本进行训练,最终实现海洋微生物种群多样性信息自适应分类。实验结果表明:所提方法的平均精度均值(mean Average Precision,mAP)稳定在95%以上,可以有效提升海洋微生物种群多样性信息自适应分类的准确性和分类效率。In order to achieve accurate and efficient adaptive classification of marine microbial population diversity information,a SVM algorithm based adaptive classification method for marine microbial population diversity information is proposed.Deploying marine microbial population information collection points through wireless sensor networks,establishing a channel model for marine microbial population information collection,designing carrier frequency encoding under distance constrained bandwidth,and utilizing the Porter interval balance control method to achieve the optimization of marine microbial population information collection.In the kernel space,spectral clustering is performed on the collected marine microbial population information to determine representative information points.At the same time,SVM is introduced to train all samples,ultimately achieving adaptive classification of marine microbial population diversity information.The experimental results show that the mAP value of the proposed method is stable above 95%,which can effectively improve the accuracy and classification efficiency of adaptive classification of marine microbial population diversity information.

关 键 词：SVM算法 海洋微生物 种群多样性信息 自适应分类 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]