报告人：Edward Rolf Tufte

题目：Practical Advice for Data Visualization, Machine Learning, and Artificial Intelligence

时间：7月23日上午08:45-09:45 青海宾馆会议中心大会堂

个人简介：Edward Tufte taught data analysis and public policy as a professor at Princeton and Yale for 31 years. He received a B.S. and M.S. in statistics from Stanford, a Ph.D. in political science from Yale, and 9 honorary degrees.Edward Tufte, via his Graphics Press, wrote, designed, and self-published 5 books on analytical thinking and showing, taught a one-day course, Presenting Data and Information for 923 days to 328,001 students, and published Artful Sentences: Syntax as Style by Virginia Tufte, his mother’s classic book.Beginning in 2006, ET designed and constructed a 234-acre art park, Hogpen Hill Farms,showing his outdoor artworks, which are also in collections at Fermilab, Two World Trade Center, and Portola Valley, California.
Website: www.edwardtufte.com
Email: edward.tufte@yale.edu

报告人：刘世霞，清华大学

题目：以数据为中心的可解释人工智能

时间：7月25日上午08:30-09:30 青海宾馆一楼迎宾厅

摘要：研究表明，在机器学习中，训练数据的质量决定了数据分析效果的上限，而模型和算法的改进只是不断逼近这个上限。因此，有效的数据质量分析和提升至关重要。本报告将介绍我们提出的基于可视分析的数据质量可解释框架。该框架将机器学习方法和可视化技术紧密集成在一起，帮助数据专家更好地理解训练数据中的错误和不一致，提高数据的可用性。最后，结合具体的应用实例，如标注数据修正和数据集偏差消除等，介绍我们基于该框架研制开发的可视分析方法与技术。

个人简介：刘世霞，清华大学教授，国家级人才计划入选者，IEEE Fellow。主要研究方向是可解释机器学习，文本可视分析和文本挖掘。2020入选IEEE Visualization Academy。担任 CCF A类会议 IEEE VIS(VAST) 2016和 2017的论文主席，IEEE VIS 2020-2023指导委员会委员；担任 IEEE Transactions on Visualization and Computer Graphics 副主编（ Associate editor-in-chief）、曾任编委；担任CCF A类期刊Artificial Intelligence编委；担任 IEEE Transactions on Big Data 和ACM Transactions on Interactive Intelligent Systems的编委。