...
  • 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Advanced Science Letters >The Comparison of Three Selection Techniques for Numerical Attribute Reduction
【24h】

The Comparison of Three Selection Techniques for Numerical Attribute Reduction

机译：数值减少三种选择技术的比较

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Metahueristic algorithms have been spotted as the potential technique in attribute reduction for efficient data classification. Despite having promising results, some of them are still unexplored in terms of feature reduction capability. To the best of our knowledge, the potential andcapability of Bat algorithm (BA) in attribute reduction have yet been reported. Thus, the objective of this paper is to highlight the potential of Bat algorithm (BA) for feature reduction. A thorough experiment has been performed on three techniques: Information Gain (IG), Discrete ParticleSwarm Optimization-Rough Set (DPSO-RS), and Bat algorithm (BA) on 14 benchmark datasets to obtain their performance based on reduction percentage. Those techniques are chosen due to different reduction strategies. The results showed that a number of attributes from nine datasets have beensuccessfully reduced by BA with 33%–88% as compared to IG and DPSO-RS. These promising results show that BA has the potential in reducing data dimensionality for an NP-hard problem.
机译：已经发现融合算法作为有效数据分类的属性降低中的潜在技术。尽管具有有希望的结果，但其中一些仍然在特征减少能力方面仍未开发。据我们所知，尚未报告蝙蝠算法（BA）在属性减少中的潜在和可容性。因此，本文的目的是突出BAT算法（BA）的潜在特征减少。已经有三种技术进行了彻底的实验：信息增益（IG），离散综合体Warm优化 - 粗糙集（DPSO-RS）和14个基准数据集的BAT算法（BA），以基于减少百分比获得其性能。这些技术是由于不同的减少策略而选择的。结果表明，与IG和DPSO-RS相比，BA的BA具有33％-88％的九个数据集的许多属性。这些有希望的结果表明，BA具有降低NP难题的数据维度的潜力。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号