查询结果:   陈超.混合Boost算法实现的行人检测技术[J].计算机应用与软件,2019,36(6):184 - 189.
中文标题
混合Boost算法实现的行人检测技术
发表栏目
人工智能与识别
摘要点击数
60
英文标题
PEDESTRIAN DETECTION TECHNOLOGY BASED ON MIXTURE BOOST ALGORITHM
作 者
陈超 Chen Chao
作者单位
内江师范学院数学与信息科学学院 四川 内江 641110     
英文单位
School of Maths and Informations Science, Neijiang Normal University, Neijiang 641110, Sichuan, China     
关键词
SBoost算法 PBoost算法 混合Boost算法 动态权重调整 非平衡的样本采样
Keywords
SBoost algorithm PBoost algorithm Mixture Boost algorithm Dynamic weight adjustment Unbalanced sample sampling
基金项目
作者资料
陈超,讲师,主研领域:数字图像处理,视频目标检测。 。
文章摘要
传统AdaBoost存在一定的局限,比如训练分类器时对训练样本自身所带的噪声过于敏感,产生的分类器泛化能力不强和导致分类器过拟化问题,在训练分类器时只能静态分配分类器权重而不能自适应地对每个训练样本动态调整权重等问题。提出一种基于SBoost算法和PBoost算法,引入样本权重调节器、非平衡的样本采样、误差纠偏方法来检测潜在的样本。模拟实验表明:改进后的技术有效的提高了分类器的精确度且防止过拟化问题。
Abstract
Traditional AdaBoost has some limitations, such as too sensitive to the noise of training samples, weak generalization ability of classifiers leading over-fitting. When training classifier, the weight of classifier can only be allocated statically, but the weight of each training sample cannot be dynamically adjusted adaptively. To solve this problem, this paper proposed a new method based on SBoost algorithm and Boost algorithm, and introduced sample weight adjuster, unbalanced sample sampling and the error correction method to detect potential samples. The simulation results show that the improved method can effectively improve the accuracy of the classifier and prevent over-fitting.
下载PDF全文   

根据该篇关键词查找到本刊已发表相关论文供参考
序号
文  章  标  题
作者1
发表栏目
页码
摘要
1
混合Boost算法实现的行人检测技术
陈超
人工智能与识别
2019
6
184
[摘要]