特邀报告

张钹

张钹

教授,中国科学院院士

清华大学计算机系

报告题目

我与人工智能

摘要

1978年清华大学计算机系成立了“人工智能与智能控制”的研究方向,人工智能的教学与科研从此开始。1990年2月“智能技术与系统”国家重点实验室正式对外开放运行。我们为什么从事人工智能研究,又如何从事人工智能研究?30多年中,我进行哪些研究工作,取得哪些成果,有什么经验与体会。

在人工智能人才培养上,我是如何培养研究生,有什么经验与体会。

我的人生理想又是什么?

报告人简介

清华大学计算机系教授,中科院院士。1958年毕业于清华大学自动控制系,同年留校任教至今。1980,2-1982,2美国伊利诺斯大学访问学者。2011年汉堡大学授予自然科学荣誉博士。曾任校学位委员会副主任,现任微软亚洲研究院技术顾问。

他参与人工智能、人工神经网络、机器学习等理论研究,以及这些理论应用于模式识别、知识工程与机器人等技术研究。在过去30多年中,他提出问题求解的商空间理论,在商空间数学模型的基础上,提出了多粒度空间之间相互转换、综合与推理的方法。提出问题分层求解的计算复杂性分析以及降低复杂性的方法。该理论与相应的新算法已经应用于不同领域,如统计启发式搜索、路径规划的拓扑降维法、基于关系矩阵的时间规划以及多粒度信息融合等,这些新算法均能显著降低计算复杂性。该理论现已成为粒计算的主要分支之一。在人工神经网络上,他提出基于规划和基于点集覆盖的学习算法。这些自顶向下的结构学习方法比传统的自底向上的搜索方法在许多方面具有显著优越性。

Jon Atli Benediktsson

Jon Atli Benediktsson

Prof. Jón Atli Benediktsson, Fellow IEEE, Fellow SPIE

Faculty of Electrical and Computer Engineering

University of Iceland

Biography

Jon Atli Benediktsson is Pro-Rector of Academic Affairs and Professor of Electrical and Computer Engineering at the University of Iceland. He received the Cand. Sci. degree in electrical engineering from the University of Iceland, Reykjavik, Iceland, in 1984 and the M.S.E.E. and Ph.D. degrees in electrical engineering from Purdue University,West Lafayette, IN, in 1987 and 1990, respectively. His research interests are in pattern recognition, remote sensing, image analysis, biomedical analysis of signals, and signal processing, and he has published extensively in those fields. He is a cofounder of the biomedical startup company Oxymap. Prof. Benediktsson is the 2011-2012 President of the IEEE Geoscience and Remote Sensing Society (GRSS) and has been on the GRSS Administrative Committee since 2000. He was the Editor of the IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING (TGRS) from 2003 to 2008 and has served as an Associate Editor of TGRS since 1999 and the IEEE GEOSCIENCE and REMOTE SENSING LETTERS since 2003.He received the Stevan J. Kristof Award from Purdue University in 1991 as outstanding graduate student in remote sensing. In 1997, he was the recipient of the Icelandic Research Council’s Outstanding Young Researcher Award; in 2000, he was granted the IEEE Third Millennium Medal;in 2004, he was a corecipient of the University of Iceland’s Technology Innovation Award; in 2006, he received the yearly research award from the Engineering Research Institute of the University of Iceland; and in 2007, he received the Outstanding Service Award from the IEEE Geoscience and Remote Sensing Society. He is co-recipient of the 2012 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING Best Paper Award, and a co-recipient of the 2013 IEEE GRSS High Impact Paper Award. In 2013 he received the IEEE/VFI Electrical Engineer of the Year Award. Prof. Benediktsson is a Fellow of SPIE. He is a member of Societas Scientiarum Islandica and Tau Beta Pi.

Chris H.Q. Ding (丁宏强)

University of Texas at Arlington

报告题目

Sparse Coding and Low-rank Matrix Models for Data Representation and Feature Selection

Abstract

Sparse coding and low rank models are mostly based on matrix models using L1 norm, L21 norm, trace norm, etc. Sparse coding starts from LASSO using L1 norm for 2-class feature selection, and grow into L21 norm based multi-class feature selection. Group LASSO incorporates class label information. L2p norm based models show further improvements. Dictionary learning obtains data representations better than PCA by learning the basis and codes with L1 regularization. Robust dictionary learning uses L21 norm or L1 norm for data representation. Trace norm is used to enforce low rank in data representation.Many new ideas and variants are proposed. In this talk, we survey these new and growing areas, and their main computational methods: proximal gradient descent, iteratively reweighted (EM-like) method, and augmented Lagrangian method. We note that although convex formulations are mathematically elegant, their solutions are unique, and are heavily emphasized here, non-convex formulations could be simpler and easier to understand (including model parameters), faster to solve, and produce better results for real applications.

报告人简介

丁宏强在美国哥伦比亚大学李政道教授研究小组求学,获得哥伦比亚大学博士学位。长期工作于美国加州理工学院,美国喷气动力实验室及美国劳伦斯-伯克利国家实验室。2007年加入德州大学阿灵顿分校任教授。他的研究领域包括数据挖掘,机器学习,信息检索,高性能计算等。从 2001 年开始,他和合作者创立了用矩阵模型作为中心理论和计算方法的子领域,研究PCA和K均值聚类的等价性,非负矩阵分解的聚类特性,提出矩阵L21范数的概念。发表200余篇论文,被引用13400 次。曾在加州大学伯克利分校、斯坦福大学、卡耐基梅隆大学、滑铁卢大学、阿尔伯塔大学、Google研究院、IBM研究院、Microsoft研究院、香港大学、香港科技大学、新加坡国立大学、北大、清华做学术报告。多次担任美国,香港,爱尔兰,以色列等国家科学基金会项目评审人。获得四篇最佳论文奖。第三批国家千人计划,安徽大学特聘教授。

周志华

南京大学

报告题目

从AdaBoost到大间隔分布学习机

摘要

本报告先介绍关于AdaBoost最近的理论研究结果,然后介绍在该结果启示下产生的机器学习算法设计新思想,以及随之而来的有效算法。

报告人简介

周志华,南京大学教授,ACM Distinguished Scientist,IEEE Fellow,IAPR Fellow,中国计算机学会会士。国家杰出青年基金获得者,长江学者特聘教授。主要从事机器学习、数据挖掘、模式识别等领域的研究工作,在领域内一流国际期刊和顶级国际会议发表论文逾百篇,论文被引用万余次,获发明专利13项。先后担任十六种SCI(E)期刊的执行主编、副主编、编委等。三十余次国际会议主席或领域主席,亚洲机器学习会议发起人。曾获国家自然科学二等奖、两次教育部自然科学一等奖、IEEE计算智能杰出青年成就奖、中国青年科技奖、霍英东一等奖、微软青年教授奖、12次国际期刊/会议论文、报告或竞赛奖等。