4.txt

This complete matlab for neural network

发信人: GzLi (笑梨), 信区: DataMining
标  题: [转载] Talk 12.26
发信站: 南京大学小百合站 (Mon Dec 23 12:39:32 2002)

【 以下文字转载自 AI 讨论区 】
【 原文由 daniel 所发表 】

  12月26日  10:00  逸夫馆605

  题目：Pattern Similarity and Discovery

  报告人：Dr. Zhongfei Zhang

  单位：SUNY Binhamton

  报告人简历：

  Zhongfei Zhang is an assistant professor at the Computer Science Department

  at State University of New York (SUNY) at Binghamton. He received B.S. in

  Electronics Engineering (with Honors), M.S. in Information Sciences, both

  from Zhejiang University, China, and PhD in Computer Science from the

  University of Massachusetts at Amherst. He was on the faculty of Computer

  Science and Engineering Department, and a research scientist at the

  Center of Excellence for Document Analysis and Recognition (CEDAR), both

  at SUNY Buffalo, before he joined the faculty of SUNY Binghamton. His

  research interests include Computer Vision and Image Understanding,

  Pattern Recognition, Data Mining, Information Fusion, and Multimedia

  Information Indexing and Retrieval, as well as Biomedical Engineering.

  He has been PIs/Co-PIs for several projects in these areas supported by
 
 the federal government, the NY State government, as well as private industries.

  He has three inventions, has served as reviewers/PC members for many

  conferences and journals, as grant review panelists for governmental

  and private funding agencies, and holds one journal editorial board

  position. He has also served as technical consultants for AAI, GE, DVC,

  and Universal Instruments.

  报告摘要：

  In this talk I will overview three active research projects I am directing

  in the Multimedia Research Lab at Binghamton University. That’s why the 

  title of this talk is so broad and generic. The first project focuses on 

  effective and efficient image retrieval from a large collection of imagery 

  data, in which we have developed a fuzzy logic based, semantics-sensitive, 

  and efficient indexing and retrieval methodology, called FUSE. FUSE aims at 

  addressing the three issues identified in the current literature of content 

  based image retrieval: the robustness of the indexing scheme, the efficiency 

  of the retrieval especially from a large collection of imagery data, and the 

  subjectivity of the retrieval. Solutions to these issues are developed in 

  FUSE methodology. The second project focuses on independent motion detection 

  from large collection of surveillance data. In this project, we have 

  developed a qualitative approach that can be applied directly to compressed 

  MPEG video streams to detect those shots that contain independently moving 

  objects. The approach is called QLS, and is based on the consistency theory 

  of a linear system. QLS has been recognized as a significant step towards 

  advancing the suveillance technology and capability in many applications. 

  The third project focuses on automatic mining and pattern discovery from 

  large collection of textual data. In particular, the project targets on the 

  topic of money laundering cases in law enforcement applications. Crime group 

  models are automatically generated based on correlation analysis along the 

  timeline, and consequently we call the prototype system as CORAL. The CORAL 

  technology has showcased a significant saving in government crime 

  investigation effort, and has proved to hold great promise in further 

  development. If time allows, I will show the demos of all the three projects.

--
宠辱不惊  闲看庭前花开花落
去留无意  漫随天外云卷云舒

※ 来源:．南京大学小百合站 bbs.nju.edu.cn．[FROM: 202.119.48.227]
--
※ 转载:．南京大学小百合站 bbs.nju.edu.cn．[FROM: 211.80.38.17]