基于主动学习的数据流分类器分析-计算机系统结构专业论文.docxVIP

重庆大学硕士学位 重庆大学硕士学位论文 英文摘要 PAGE PAGE VI classification confidence should be submitted to human experts. DCAs availability is demonstrated by the timing analysis and its correctness is confirmed through the experiments on emulational data set and real data set. Thirdly, the paper presents a classifier called RCL which is based on Storm. In real environment many applications need to classify large amount of data quickly. Its hard and not necessary to design a distributed classifier all by yourself. A design based on Hadoop is a alternative but not the best choice. Eventually we choose the design based on Storm after some survey. Storm is a free and open source distributed realtime computing system. RCL applies DCA on top of Storm to get classifier that can process huge volumes of data quickly. Through the experiments on on emulational data set and real data set, the availability and correctness of RCL is demonstrated by its speed-up, parallel efficiency and average accuracy. Keywords: Data Streams, Classification, Concept Drifting, Active Mining, Storm 重庆大学硕士学位论 重庆大学硕士学位论文 目 录 目 录 中文摘要 ···································································································· I 英文摘要 ··································································································· II HYPERLINK \l _bookmark0 1 绪论 1 HYPERLINK \l _bookmark1 1.1 研究背景及意义 ·······················································································1 HYPERLINK \l _bookmark2 1.1.1 研究背景 ··························································································1 HYPERLINK \l _bookmark3 1.1.2 研究意义 ··························································································2 HYPERLINK \l _bookmark4 1.2 国内外研究现状 ·······················································································2 HYPERLINK \l _bookmark5 1.2.1 单分