VALSE

VALSE 首页 活动通知 查看内容

20210602-14 总第240期 Deep AUC Maximization

2021-5-28 11:59| 发布者: 程一-计算所| 查看: 2087| 评论: 0

摘要: 报告时间2021年06月02日 (星期三)晚上21:00 (北京时间)主 题Deep AUC Maximization: Algorithms and Applications主持人徐易 (Alibaba)报告嘉宾:刘名睿 (Boston University)报告题目:Fast Algorithms for AUC Maxi ...

报告时间

2021年06月02日 (星期三)

晚上21:00 (北京时间)

主  题

Deep AUC Maximization: Algorithms and Applications

主持人

徐易 (Alibaba)


报告嘉宾:刘名睿 (Boston University)

报告题目:Fast Algorithms for AUC Maximization


报告嘉宾:袁卓宁 (University of Iowa)

报告题目:How to Win Chexpert Competition? Deep AUC Maximization in Medical Applications



Panel嘉宾:

Tianbao Yang (University of Iowa)、Yiming Ying (State University of New York at Albany)、刘名睿 (Boston University)、袁卓宁 (University of Iowa)


Panel议题:

1. How can we optimize AUPRC (Areas Under Precision-Recall Curves)?

2. How can we perform end-to-end training for deep AUC maximization?

3. Can we use adaptive algorithms (such as Adam) for optimizing the objectives for AUC and do they have convergence?

4. Do we need to use the two-stage methods for deep AUC maximization?

5. What other research directions or questions about deep AUC maximization?


*欢迎大家在下方留言提出主题相关问题,主持人和panel嘉宾会从中选择若干热度高的问题加入panel议题!


报告嘉宾:刘名睿 (Boston University)

报告时间:2021年06月02日 (星期三)晚上21:00 (北京时间)

报告题目:Fast Algorithms for AUC Maximization


报告人简介:

Mingrui Liu is a postdoctoral fellow at Rafik B. Hariri Institute for Computing at Boston University, working with Francesco Orabona. Previously he obtained his Ph.D. at The University of Iowa in August 2020 under the advise of Tianbao Yang. He has also spent time working as a research intern at IBM T. J. Watson Research Center. His research interests include machine learning, optimization, deep learning and lifelong learning.


个人主页:

https://mingrliu.github.io


报告摘要:

Area Under the ROC Curve (AUC) is an important metric in machine learning, especially in the imbalanced data setting. In this talk, I will focus on provably efficient algorithms for online AUC maximization with both linear model and deep neural network to learn from imbalanced data. The main result is that AUC maximization can be reformulated as a min-max objective and algorithms with fast convergence rates are established.


参考文献:

[1] Fast Stochastic AUC Maximization with O(1/n) Convergence Rate. Mingrui Liu, Xiaoxuan Zhang, Zaiyi Chen, Xiaoyu Wang, Tianbao Yang. ICML 2018.

[2] Stochastic AUC Maximization with Deep Neural Networks. Mingrui Liu, Zhuoning Yuan, Yiming Ying, Tianbao Yang. ICLR 2020.


报告嘉宾:袁卓宁 (University of Iowa)

报告时间:2021年06月02日 (星期三)晚上21:30 (北京时间)

报告题目:How to Win Chexpert Competition? Deep AUC Maximization in Medical Applications


报告人简介:

Zhuoning Yuan is a PhD student at the Department of Computer Science at the University of Iowa. His research interests lie primarily in developing robust machine learning solutions to improve disease diagnosis for imbalanced medical data. He has published several papers in top-tier AI conferences, including ICML, NeurIPS, ICLR, KDD. He received 1st place in CheXpert competition in detecting chest and lung-related diseases for X-Ray images. He served as reviewers/PC members for several top-tier conferences/journals and he also received the Outstanding Reviewer Award in CVPR 2021.


个人主页:

https://homepage.divms.uiowa.edu/~zhuoning/


报告摘要:

In this talk, I will introduce a new AUC loss function and talk about Deep AUC Maximization in two medical applications, including classification of chest x-ray images for identifying chest and lung-related diseases and classification of images of skin cancer for identifying melanoma.


参考文献:

[1] Yuan, Z., Yan, Y., Sonka, M., & Yang, T. Robust Deep AUC Maximization: A New Surrogate Loss and Empirical Studies on Medical Image Classification. arXiv preprint arXiv: 2012.03173. 2020.

[2] Liu, M., Yuan, Z., Ying, Y., & Yang, T. Stochastic AUC maximization with deep neural networks. ICLR 2020.

[3] Yuan, Z., Guo, Z., Xu, Y., Ying, Y., & Yang, T. Federated Deep AUC Maximization for Heterogeneous Data with a Constant Communication Complexity. ICML 2021.


Panel嘉宾:Tianbao Yang (University of Iowa)


嘉宾简介:

Tianbao Yang is an associate professor at the Computer Science Department at the University of Iowa. He received his Ph.D. degree in Computer Science from Michigan State University in 2012. Prof. Yang’s research interests center around optimization, big data, machine learning and AI. He has published more than 100 papers including NeurIPS, ICML, ICLR, COLT, AAAI, IJCAI, KDD, CVPR, ECCV, JMLR, and ML. Prof. Yang won the Best Student Paper Award at COLT 2012 and received NSF Career Award in 2019. His group recently received the 1st place at Stanford CheXpert Competition. He has served as Area Chair/PC member/Reviewer/Associate Editor for several top machine learning and deep learning conferences and journals.


个人主页:

https://homepage.cs.uiowa.edu/~tyng/index.html


Panel嘉宾:Yiming Ying (State University of New York at Albany)


嘉宾简介:

Yiming Ying is a Professor at the Department of Mathematics and Statistics, SUNY Albany (UAlbany) and currently leading the machine learning group, ML@UA, at UAlbany. His research interests center on Machine Learning, Statistical Learning Theory, and Large-Scale Optimization. He is currently working on research topics of deep learning for imbalanced classification, and robust and trustworthy machine learning. He has served as an associate editor of Neurocomputing, Mathematical Foundation of Computing, and Mathematics of Computation and Data Science, as a Senior PC member or Area Chair for AISTATS and NeurIPS, and as a PC member for ICML, IJCAI, AAAI, and COLT. To date, he has published more than 70 papers in top-tier machine learning conferences/journals and leading applied math journals.


个人主页:

https://www.albany.edu/~yy298919/


主持人:徐易 (Alibaba)


主持人简介:

Yi Xu is a Senior Algorithm Engineer at Machine Intelligence Technology Team, Alibaba Group (U.S.) Inc. He received his Ph.D. degree in Computer Science from the University of Iowa in 2019. Dr. Xu’s research interests focus on optimization, machine learning, and deep learning. He has published 17 papers including NeurIPS, ICML, AAAI, IJCAI, and UAI. He has served as PC member or Reviewer for several top machine learning and deep learning conferences and journals.


个人主页:

https://yxu71.github.io




21-14期VALSE在线学术报告参与方式:

长按或扫描下方二维码,关注“VALSE”微信公众号 (valse_wechat),后台回复“14期”,获取直播地址。


特别鸣谢本次Webinar主要组织者:

主办AC:徐易 (Alibaba)

责任AC:王楠楠 (西安电子科技大学)



活动参与方式

1、VALSE Webinar活动依托在线直播平台进行,活动时讲者会上传PPT或共享屏幕,听众可以看到Slides,听到讲者的语音,并通过聊天功能与讲者交互;


2、为参加活动,请关注VALSE微信公众号:valse_wechat 或加入VALSE QQ群(目前A、B、C、D、E、F、G、H、I、J、K、L、M、N群已满,除讲者等嘉宾外,只能申请加入VALSE Q群,群号:698303207);

*注:申请加入VALSE QQ群时需验证姓名、单位和身份,缺一不可。入群后,请实名,姓名身份单位。身份:学校及科研单位人员T;企业研发I;博士D;硕士M。


3、在活动开始前5分钟左右,讲者会开启直播,听众点击直播链接即可参加活动,支持安装Windows系统的电脑、MAC电脑、手机等设备;


4、活动过程中,请不要说无关话语,以免影响活动正常进行;


5、活动过程中,如出现听不到或看不到视频等问题,建议退出再重新进入,一般都能解决问题;


6、建议务必在速度较快的网络上参加活动,优先采用有线网络连接;


7、VALSE微信公众号会在每周四发布下一周Webinar报告的通知及直播链接。


8、Webinar报告的PPT(经讲者允许后),会在VALSE官网每期报告通知的最下方更新[slides]


9、Webinar报告的视频(经讲者允许后),会更新在VALSEB站、西瓜视频,请在搜索Valse Webinar进行观看。


刘名睿 [slides]

袁卓宁 [slides]


小黑屋|手机版|Archiver|Vision And Learning SEminar

GMT+8, 2024-11-10 06:12 , Processed in 0.018302 second(s), 14 queries .

Powered by Discuz! X3.4

Copyright © 2001-2020, Tencent Cloud.

返回顶部