|
| 一种基于MapReduce的动态数据流分类算法 |
| A dynamic data stream classification algorithm based |
| 投稿时间：2013-06-14 修订日期：2014-01-08 |
| DOI： |
| 中文关键词: 数据流分类 增量式学习 ESVM MapReduce 遗忘因子 鲁棒 |
| 英文关键词: data stream classification incremental learning ESVM MapReduce forgetting factor robust |
| 基金项目:国家自然科学基金项目（61173163, 51105052)，新世纪人才（NCET-09-0251），辽宁省教育厅科研项目（201102037） |
|
| 摘要点击次数: 1449 |
| 全文下载次数: 0 |
| 中文摘要: |
| 当前动态数据流下的实时分类问题存在三个难点1：针对海量数据的实时处理，2、概念漂移的跟踪和模型的更新，3、模型的稳定和鲁棒性。针对所述问题，将极端支持向量机(Extreme Support Vector Machine, ESVM)与MapReduce框架结合，提出了带遗忘因子的带遗忘因子的鲁棒ESVM算法，该方法通过构造残差权矩阵，对残差进行修正，同时加入遗忘因子，提高新样本的作用.实现对大数据的实时处理。实验结果表明，该改进的方法可以快速有效的对动态数据流进行分类，且结果不易受到噪声干扰，稳定性强。 |
| 英文摘要: |
| There are three difficulties in Real-time dynamic data stream classification: 1 Real-time processing of massive data. 2 Tracking of concept drift and model updates, 3 Model’s stability and robustness. To solve this problem, we combine extreme support vector machine (ESVM) with MapReduce framework, propose a forgetting factor robust ESVM algorithm (FFR-ESVM). The proposed algorithm amend the residuals by constructing a residual matrix, while improve the effect of new samples by forgetting factor. Experimental results show that the proposed algorithm can be rapid and effective for dynamic data stream classification, and the results are stable and less affected by noise interference. |
|
View Fulltext
查看/发表评论 下载PDF阅读器 |
| 关闭 |