专业建站系统 - 打造最好的网站内容系统!

http://www.zbzzjsqk.com

当前位置: 主页 > 计算机应用 > 王忠民,王斌/多频段时域分解的行为识别特征优选方法

王忠民,王斌/多频段时域分解的行为识别特征优选方法

时间:2015-07-30来源:未知 作者:admin点击:
多频段时域分解的行为识别特征优选方法 王忠民,王斌 (西安邮电大学 计算机学院,西安 710121) 摘要: 为了降低外界环境对移动用户行为识别的影响,保留行为敏感特征、提高行为识别的准确率,提出了一种多频段时域分解的人体行为识别特征优选方法。该方法
多频段时域分解的行为识别特征优选方法

王忠民,王斌
(西安邮电大学 计算机学院,西安 710121)

 
摘要:为了降低外界环境对移动用户行为识别的影响,保留行为敏感特征、提高行为识别的准确率,提出了一种多频段时域分解的人体行为识别特征优选方法。该方法对行为样本数据进行多频段分解,计算样本数据在不同频段信号的特征,利用遗传算法以决策树作为分类器进行特征优选,在多组特征中搜索出近似最优的特征组合。实验结果表明,该方法优选出的特征组合能有效提高行为识别的准确率。
关键词:多频段时域分解;行为识别;特征优选;遗传算法;决策树

Behavior recognition method for multi band time domain decomposition

Wang Bin, Wang Zhongmin

(Computer college, Xi'an University of Posts and telecommunications, Xi'an 710121)

Abstract: in order to reduce the influence of external environment of mobile user behavior identification, retention behavior sensitive characteristics, improve behavior recognition accuracy, human behavior recognition feature optimization method for a multi frequency time domain decomposition is proposed. The method of sample data were multi band decomposition, calculate sample data in the characteristics of different frequency band signals, by decision tree as a classifier for feature selection using genetic algorithm, in the multi group feature search approximate optimal feature combination. Experimental results show that the proposed method can effectively improve the accuracy of behavior recognition.
Keywords: multi frequency domain decomposition; behavior recognition; feature selection; genetic algorithm; decision tree

顶一下
(0)
0%
踩一下
(1)
100%
------分隔线----------------------------