当前位置>>首页 >> 分会场报告人
李新德 教授
2021/8/3 16:12:00    

李新德 教授



李新德,博士,教授,博士生导师,俄罗斯自然科学院院士,中国人工智能学会智能机器人专委会副主任委员,中国人工智能学会智能产品与产业工作委员会副主任委员等。主要研究方向:智能信息处理、无人系统、机器视觉、机器感知、智能机器人、人机交互等。新加坡国立大学ECE系博士后,美国佐治亚理工大学国家公派访问学者。承担包括863重点、国家自然科学基金重大研究计划项目、面上项目、十三五预研重点项目、JKW163重点项目、JKW重大专项等国家级项目10+项、省部级项目8项,其它项目20+项。在IEEE汇刊TIE、TII、 TFS、TM等国内外核心期刊和会议发表SCI、EI收录的论文80余篇,3篇Book Chapter,2部著作,授权国家发明专利17项,软件著作权9个。获国际科学贡献奖、中国自动化学会科技进步一等奖、省自然科学三等奖、人工智能学会最佳青年科技成果奖、 十二五航空基金优秀成果奖等各一项。

报告题目:Multi-Granularity Fusion Based on Belief Function Theory
        
报告摘要:With the advent of the big data era, the multi-source, heterogeneous and rapid evolution of data have brought huge challenges to traditional fusion decision-making. In this talk, I will introduce the novel paradigm to solve the problem of multi-source approximate fusion: multi-granularity fusion based on the belief function theory. First, An in-depth study on the multi-granularity representation of knowledge will be given. Taking the application of activity recognition as an example, I will show how to achieve the accurate modeling of multi-granular activities based on the concepts of the focal elements; Then, considering the difficulties of computing the basic belief assignments of focal elements, two different strategies: generative and discriminant will also presented. Finally, the decisions can be made through the classical multi-granularity fusion rules.