首页 | 本学科首页   官方微博 | 高级检索  
     检索      

一种改进的Adaboost训练算法
引用本文:李文辉,倪洪印.一种改进的Adaboost训练算法[J].吉林大学学报(理学版),2011,49(3):498-504.
作者姓名:李文辉  倪洪印
作者单位:吉林大学 计算机科学与技术学院, 长春 130012
基金项目:国家自然科学基金,国家高技术研究发展计划863项目基金,吉林省科技发展计划项目
摘    要:针对传统的Adaboost训练算法在训练过程中可能出现训练退化和训练目标类权重分布过适应的问题, 提出一种改进的Adaboost训练算法. 改进算法通过调整加权误差分布限制目标类权重的扩张, 并且最终分类器输出形式以概率值输出代替传统的离散值输出, 提高了训练结果的检测率. 实验结果表明, 改进的Adaboost算法在Inria数据集上取得了较好效果.

关 键 词:误差分布    Adaboost算法    权重更新    正负误差比  分类器输出  
收稿时间:2010-05-21

An Improved Adaboost Training Algorithm
LI Wen-hui,NI Hong-yin.An Improved Adaboost Training Algorithm[J].Journal of Jilin University: Sci Ed,2011,49(3):498-504.
Authors:LI Wen-hui  NI Hong-yin
Institution:College of Computer Science and Technology, Jilin University, Changchun 130012, China
Abstract:In view of the problem of degradation issues as well as the distribution of target class weights adapted to the phenomenon that may arise in the training process of the traditional Adaboost algorithm, the authors introduced a few improved methods to these problems. The article presented a modified Adaboost algorithm based on the adjusted weighted error distribution to limit the expansion weights. In addition, the Adaboost algorithm improved the classifier output forms, i.e., using output of the probability value instead of the discrete value and increased the detection rate more dramatically. Experiment shows that the test rate of the improved Adaboost algorithm could achieve excellentresults in the Inria data set. There are good prospects of application in the field of video security surveillance.
Keywords:error distribution  Adaboost algorithm  weight update  positive and negative error ratio  classifier output  
本文献已被 CNKI 万方数据 等数据库收录!
点击此处可从《吉林大学学报(理学版)》浏览原始摘要信息
点击此处可从《吉林大学学报(理学版)》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号