冗余特征检测与利用技术的研究

项目名称： 冗余特征检测与利用技术的研究

项目编号： No.60873129

项目类型： 面上项目

立项/批准年度： 2009

项目学科： 金属学与金属工艺

项目作者： 李国正

作者单位： 同济大学

项目金额： 30万元

中文摘要： 对照项目计划书，项目组高质量全面完成了拟定的研究内容。项目组在高水平论文发表、国际学术合作交流、人才培养等方面均超过项目申请书的预期成果要求。 研究内容包括三个方面：1) 研究了冗余特征检测新技术，结合多类、多标记、半监督数据特点提出了多种冗余特征检测算法；2) 研究了冗余特征重用新技术，提出了ELEC、APLSC等结合特征抽取的新分类算法；3)结合具体科学领域数据的特点，将本项目所提出的新技术，应用到中医数据分析，蛋白质亚细胞定位和信息安全等科学问题上。 项目资助下，在国际会议上报告18篇次。在IEEE T. NanoBioscience, PRJ，PRL，NCA，BMCCAM，中国科学等期刊接受和发表论文34篇，其中发表的论文中SCI收录11篇，接受论文中5篇SCI源刊。受邀为国外著作撰写3章次。先后参与组织了IEEE和ACM的两个国际会议，连续两年组织主持了中医信息学研讨会，参编4本国际会议论文集。项目成果2010年获得第一届上海中医药科技一等奖，2011年获得国际信息安全竞赛（CDMC2011）第一名.

中文关键词： 特征选择;冗余特征;特征抽取;模式识别;生物医学

英文摘要： Compared to the project plan,our team completed the project in overal high-quality. Papers published in high-level publishments, international academic cooperation and exchange, post-students education, etc.are beyond the expected requirements of the projectapplication. The study includes three aspects:1) to study the redundancy feature detection technology, combined with the characteristics of multiclass, multilabel, semi-supervised data, we invent a variety of feature detection algorithm, which effectively reduce the feature dimension; 2) to study the reuse of redundant features, new technologies,including ELEC, APLSC are proposed via combination of the new feature extraction algorithm to improve the classification results; 3) combined with the characteristicsof scientific data, we apply the new technologies proposed in this project to data analysis of Chinese medicine, proteinsub-cellular localizationand information security issues,to promote the progress of related topics. Funded by this project, we report 18 works on international conferences. IEEE T. NanoBioscience, PRJ, PRL, BMCCAM,Science in China journals and etc accepted and published 34 papers, including 11 SCI indexed papers, and 5 SCI indexed source journals.We has participated in the organizationoftwo IEEE and ACM International Conference, presided over two consecutive years workshops on Chinese medicine informatics. We edited four international conference proceedings published by ACM and IEEE. Achievements of the project received the first prize of Shanghai Chinese Medical Society in 2010, and the champion of the international information security competition (CDMC2011) in 2011.

英文关键词： Feature selection; redundant features; feature extraction; pattern recognition; biomedical data