OPT OpenIR  > 光谱成像技术研究室
Feature Combination via Clustering
Hou, Jian1,2; Gao, Huijun3; Li, Xuelong4; Hou, J (reprint author), Bohai Univ, Coll Engn, Jinzhou 121013, Peoples R China.; Hou, J (reprint author), Univ Ca Foscari Venezia, European Ctr Living Technol, I-30123 Venice, Italy.
作者部门光学影像学习与分析中心
2018-04-01
发表期刊IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
ISSN2162-237X
卷号29期号:4页码:896-907
产权排序4
摘要

In image classification, feature combination is often used to combine the merits of multiple complementary features and improve the classification accuracy compared with one single feature. Existing feature combination algorithms, e.g., multiple kernel learning, usually determine the weights of features based on the optimization with respect to some classifier-dependent objective function. These algorithms are often computationally expensive, and in some cases are found to perform no better than simple baselines. In this paper, we solve the feature combination problem from a totally different perspective. Our algorithm is based on the simple idea of combining only base kernels suitable to be combined. Since the very aim of feature combination is to obtain the highest possible classification accuracy, we measure the combination suitableness of two base kernels by the maximum possible cross-validation accuracy of their combined kernel. By regarding the pairwise suitableness as the kernel adjacency, we obtain a weighted graph of all base kernels and find that the base kernels suitable to be combined correspond to a cluster in the graph. We then use the dominant sets algorithm to find the cluster and determine the weights of base kernels automatically. In this way, we transform the kernel combination problem into a clustering one. Our algorithm can be implemented in parallel easily and the running time can be adjusted based on available memory to a large extent. In experiments on several data sets, our algorithm generates comparable classification accuracy with the state of the art.

文章类型Article
关键词Feature Combination Image Classification Clustering Dominant Set
学科领域Computer Science, Artificial Intelligence
WOS标题词Science & Technology ; Technology
DOI10.1109/TNNLS.2016.2645883
收录类别SCI ; EI
关键词[WOS]TEXTURE CLASSIFICATION ; IMAGE FEATURES ; DOMINANT SETS ; KERNEL ; SCALE ; DESCRIPTORS ; RECOGNITION
语种英语
WOS研究方向Computer Science ; Engineering
WOS类目Computer Science, Artificial Intelligence ; Computer Science, Hardware & Architecture ; Computer Science, Theory & Methods ; Engineering, Electrical & Electronic
WOS记录号WOS:000427859600011
EI入藏号20170603325513
引用统计
被引频次:5[WOS]   [WOS记录]     [WOS相关记录]
文献类型期刊论文
条目标识符http://ir.opt.ac.cn/handle/181661/30015
专题光谱成像技术研究室
通讯作者Hou, J (reprint author), Bohai Univ, Coll Engn, Jinzhou 121013, Peoples R China.; Hou, J (reprint author), Univ Ca Foscari Venezia, European Ctr Living Technol, I-30123 Venice, Italy.
作者单位1.Bohai Univ, Coll Engn, Jinzhou 121013, Peoples R China
2.Univ Ca Foscari Venezia, European Ctr Living Technol, I-30123 Venice, Italy
3.Harbin Inst Technol, Res Inst Intelligent Control & Syst, Harbin 150001, Heilongjiang, Peoples R China
4.Chinese Acad Sci, Xian Inst Opt & Precis Mech, Ctr OPTical IMagery Anal & Learning OPTIMAL, State Key Lab Transient Opt & Photon, Xian 710119, Shaanxi, Peoples R China
推荐引用方式
GB/T 7714
Hou, Jian,Gao, Huijun,Li, Xuelong,et al. Feature Combination via Clustering[J]. IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS,2018,29(4):896-907.
APA Hou, Jian,Gao, Huijun,Li, Xuelong,Hou, J ,&Hou, J .(2018).Feature Combination via Clustering.IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS,29(4),896-907.
MLA Hou, Jian,et al."Feature Combination via Clustering".IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS 29.4(2018):896-907.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
Feature Combination (1581KB)期刊论文作者接受稿限制开放CC BY-NC-SA请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[Hou, Jian]的文章
[Gao, Huijun]的文章
[Li, Xuelong]的文章
百度学术
百度学术中相似的文章
[Hou, Jian]的文章
[Gao, Huijun]的文章
[Li, Xuelong]的文章
必应学术
必应学术中相似的文章
[Hou, Jian]的文章
[Gao, Huijun]的文章
[Li, Xuelong]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。