标题 | 粗糙集数据的高精度分类算法研究 |
范文 | 杨艳丽 摘 ?要: 针对传统高精度分类算法在面对不定因子时,无法确定计算数据信噪度,造成计算精度不佳的问题,提出基于属性约简的粗糙集数据的高精度分类算法。通过对影响粗糙集数据分类精度的各影响因素进行详细分析,对粗糙集数据属性进行约简,抵消对应不定因子以及信噪数据,提高粗糙集数据分类精度。实验结果表明,采用改进分类算法相比传统分类方法,其分类精度及抗噪性均有提高,且其记录结果数据致盲率较低,具有一定优势。 关键词: 粗糙集数据; 高精度分类算法; 属性约简; 属性集; 数据集; 抗噪性 中图分类号: TN911?34; TP393 ? ? ? ? ? ? ? ? 文献标识码: A ? ? ? ? ? ? ? ? ? ?文章编号: 1004?373X(2018)10?0154?03 Abstract: In allusion to the poor calculation accuracy problem caused by inability to determine the signal?to?noise degree of calculated data when uncertain factors are met in the traditional high?precision classification algorithm, a high?precision classification algorithm based on attribute reduction is proposed for rough set data. The attributes of rough set data are reduced by detailedly analyzing various factors affecting the classification accuracy of rough set data to counteract the corresponding uncertain factors and signal?to?noise data, and improve the classification accuracy of rough set data. The experimental results show that in comparison with the traditional classification method, the improved classification algorithm has certain advantages in that it has higher classification accuracy and noise immunity, and the blind rate of the recorded result data is low. Keywords: rough set data; high accuracy classification algorithm; attribute reduction; attribute set; data set; noise immunity 3 ?仿真试验 3.1 ?试验数据设置 试验从某数据网站上下载了数个执行粗糙数据,将执行粗糙数据进行粗糙集数据的高精度分类计算。为保证试验的准确性,需要对试验数据参数进行设定,试验数据设定结果如表2所示。 3.2 ?试验结果分析 分别从计算抗性上以及计算精度上进行对比,使用传统高精度分类算法与改进高精度分类算法进行比较,在不同的试验参数下,分别记录数据致盲过程的变化量以及在三种试验计算环境下的试验结果,见表3。 通过上述表3中数据可以看出,本设计的粗糙集数据的高精度分类计算方法在计算精准度上明显高于传统计算方法。对比不同的计算过程跟踪结果,本文计算方法更具有计算抗性。图1为两种方法计算致盲点数据变化。计算致盲点数据是描述计算流程的重要指标,计算致盲点数据与计算准确率成一定的倍数关系。计算致盲点数据分布越有规律说明计算准确率越高。通过图可以看出粗糙集数据的高精度分类计算方法的计算致盲点数据分布成规律递增的趋势,但传统高精度分类计算方法的计算致盲点数据分布杂乱无序;因此本设计的粗糙集数据的高精度分类计算方法比传统高精度分类计算方法更具准确性。 4 ?结 ?语 本设计的粗糙集数据的高精度分类计算方法导入粗糙集数据实现属性约简计算,有效地排除不定因子异己信噪数据的干扰,通过属性约简方式实现粗糙数据的高精度分类计算。希望通过本文的研究能够提升高精度分类计算方法的计算精准度。参考文献 [1] 王旭,谢沐男,邓蕾.基于粗糙集的物流资源分类方法研究[J].计算机工程与应用,2016,52(14):67?71. WANG Xu, XIE Munan, DENG Lei. Research on classification method of logistics resources based on rough set [J]. Computer engineering and applications, 2016, 52(14): 67?71. [2] 刘继宇,王强,罗朝晖,等.基于粗糙集的加權KNN数据分类算法[J].计算机科学,2015,42(10):281?286. LIU Jiyu, WANG Qiang, LUO Zhaohui, et al. Weighted KNN data classification algorithm based on rough set [J]. Computer science, 2015, 42(10): 281?286. [3] 徐久成,李涛,孙林,等.基于信噪比与邻域粗糙集的特征基因选择方法[J].数据采集与处理,2015,30(5):973?981. XU Jiucheng, LI Tao, SUN Lin, et al. Feature gene selection based on SNR and neighborhood rough set [J]. Journal of data acquisition & processing, 2015, 30(5): 973?981. [4] 熊方,张贤勇.变精度粗糙集的区域属性约简及其结构启发算法[J].计算机应用,2016,36(11):2954?2957. XIONG Fang, ZHANG Xianyong. Regional attribute reduction and their structural heuristic algorithms for variable precision rough sets [J]. Journal of computer applications, 2016, 36(11): 2954?2957. [5] MA X, WANG G, YU H. Heuristic algorithms for finding distribution reducts in probabilistic rough set model [J]. International journal of engineering science & technology, 2015, 2(3): 1?14. [6] HAN Y, LI M C, GUO X C, et al. The text classification attribute reduction algorithm based on the rough set theory [J]. Journal of Northeast Dianli University, 2016, 1(5): 145?150. [7] JIAO N. Research on knowledge reduction algorithm based on variable precision tolerance rough set theory [J]. Computer science, 2015, 5(8): 120?135. [8] LI X, ZENG W, SHI Y, et al. Resting state fMRI data classification method based on K?means algorithm optimized by rough set [J]. 2017, 8(6): 84?92. [9] KUMAR S U, INBARANI H H. A novel neighborhood rough set based classification approach for medical diagnosis [J]. Procedia computer science, 2015, 47: 351?359. [10] 庞帮艳,张艳敏.基于粗糙集的公共网络入侵检测方法研究[J].现代电子技术,2017,40(4):28?31. PANG Bangyan, ZHANG Yanmin. Research of public network intrusion detection method based on rough set theory [J]. Modern electronics technique, 2017, 40(4):28?31. |
随便看 |
|
科学优质学术资源、百科知识分享平台,免费提供知识科普、生活经验分享、中外学术论文、各类范文、学术文献、教学资料、学术期刊、会议、报纸、杂志、工具书等各类资源检索、在线阅读和软件app下载服务。