电子信息对抗技术
電子信息對抗技術
전자신식대항기술
ELECTRONIC INFORMATION WARFARE TECHNOLOGY
2014年
6期
60-63
,共4页
王社会%杨俊安%YANG Jun-an
王社會%楊俊安%YANG Jun-an
왕사회%양준안%YANG Jun-an
数据缺失%矩阵修复%KNN理论%自适应K%局部最近邻
數據缺失%矩陣脩複%KNN理論%自適應K%跼部最近鄰
수거결실%구진수복%KNN이론%자괄응K%국부최근린
data missing%matrix completion%KNN theory%K-value self-adaption%local search
数据在采集和传输过程中由于多种原因会造成矩阵残缺，因此在数据分析之前需要对残缺矩阵进行修复。常见的KNN修复方法k值选取不合理，且需在整个矩阵中搜索近邻，影响算法的修复效果。在其基础上提出了一种k值自适应的局部KNN矩阵修复方法，合理考虑了k值的选取和近邻项的搜索范围。实验证明了该方法能有效提高矩阵修复的正确率，且算法的时效性有所提高。
數據在採集和傳輸過程中由于多種原因會造成矩陣殘缺，因此在數據分析之前需要對殘缺矩陣進行脩複。常見的KNN脩複方法k值選取不閤理，且需在整箇矩陣中搜索近鄰，影響算法的脩複效果。在其基礎上提齣瞭一種k值自適應的跼部KNN矩陣脩複方法，閤理攷慮瞭k值的選取和近鄰項的搜索範圍。實驗證明瞭該方法能有效提高矩陣脩複的正確率，且算法的時效性有所提高。
수거재채집화전수과정중유우다충원인회조성구진잔결，인차재수거분석지전수요대잔결구진진행수복。상견적KNN수복방법k치선취불합리，차수재정개구진중수색근린，영향산법적수복효과。재기기출상제출료일충k치자괄응적국부KNN구진수복방법，합리고필료k치적선취화근린항적수색범위。실험증명료해방법능유효제고구진수복적정학솔，차산법적시효성유소제고。
Since the data may inevitably miss some elements in the process of information acqui-sition and transmission due to various reasons, it is necessary to full them. But the general KNN completion method is not reasonable on the selection of the value of k, and it needs to search for neighbors of the entire matrix. An advanced KNN matrix completion is presented by K-value self-adaption and local search, which reasonably considers the value of k and the search range of neighbors. The experiment indicates this method is effective, and the timeliness is also en-hanced.
