首页 > 科学研究 > 学术看板 > 正文

误差建模原理

供稿:    责任编辑:安果    时间:2018-05-15    阅读:

主讲人:孟德宇

报告人单位:西安交通大学数学与统计学院

报告时间:2018年5月15日15:30

报告地点:基础教学楼A303

个人主页或EMAIL:http://gr.xjtu.edu.cn/web/dymeng


报告摘要:传统机器学习主要关注于确定性信息的建模,而在复杂场景下,机器学习方法容易出现对数据噪音的鲁棒性问题,而该鲁棒性问题与误差函数的选择紧密相关。本次报告聚焦于如何针对包含复杂噪音数据进行误差建模的鲁棒机器学习原理。这一原理对在线视频处理、医学图像恢复等问题,已体现出个性化的应用优势,该原理亦有希望能够引导出更多有趣的机器学习相关应用与发现。

Traditional machine learning methods are sensitive to the noise that probably exists into data. Such a robustness issue is close to the choice of error reformulation. In this talk, I will introduce the principle of robust machine learning, as well as how to mathematically formulate the noisy data. As shown in our studies, the formulation of error has shown promising performance in video processing, medical image restoration, and so on.


报告人简介:西安交通大学数学与统计学院教授,博导。曾赴香港理工大学,Essex大学与卡内基梅隆大学进行学术访问与合作。共接收/发表论文80余篇,其中包括IEEE Trans论文22篇和CCF A类会议30篇。担任ICML,NIPS等会议程序委员会委员,AAAI2016高级程序委员会委员。目前主要聚焦于自步学习、误差建模、张量稀疏性等机器学习相关方向的研究。


Prof. Deyu Meng received the B.Sc., M.Sc., and Ph.D. degrees from Xian Jiaotong University, Xi- an, China, in 2001, 2004, and 2008, respectively. He is currently a Full Professor with the Institute for Information and System Sciences, Xian Jiao- tong University. From 2012 to 2014, he took his two- year sabbatical leave in Carnegie Mellon University. His current research interests include self-paced learning, noise modeling, and tensor sparsity.