• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>Conference on Signal and Data Processing of Small Targets 2002, Apr 2-4, 2002, Orlando, USA >Classifier Performance Model for the Many-Class Case
【24h】

Classifier Performance Model for the Many-Class Case

机译：多类案例的分类器性能模型

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

A model is presented for predicting classification performance for systems having a large population of classes. The cases of large and small training set size for each class are treated separately. A method is proposed for measuring classification performance as the mean ranking statistic ρ_E which is derived from the average information content h_E of the system feature vector, which is in turn derived from the system covariance matrices (∑_W, ∑_B)- This method for predicting ρ_E is applied to the large training set case , explaining why performance is not compromised but improved by adding noisy features. The method is extended for predicting performance in the more difficult small training set case , explaining why performance may be compromised by the addition of noisy features in that situation.
机译：提出了一种用于预测具有大量类的系统的分类性能的模型。每个班级的大型和小型训练集大小的情况将分别处理。提出了一种用于测量分类性能的方法，该方法是从系统特征向量的平均信息内容h_E得出的平均排名统计量ρ_E中得出的，该平均值又从系统协方差矩阵（∑_W，∑_B）中得出。预测ρ_E应用于大型训练集的情况，解释了为什么不通过添加噪声特征来降低性能但提高性能的原因。扩展了该方法以预测更困难的小型训练集情况下的性能，从而解释了为什么在那种情况下添加噪声特征可能会损害性能。

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