Panpan Xu (徐盼盼), Lead Research Scientist at Bosch Research, Sunnyvale, CA.
My research interest is empowering people to understand data and communicate their insights more efficiently through the development of new interactive visualization systems and data mining algorithms. I use a multidisciplinary approach combining techniques in machine learning, data visualization, and human-computer interaction.
Currently, I am working on the following topics:
- visual summarization techniques for distilling knowledge from large-scale and complex data (e.g. event sequence data, spatial-temporal data, graphs/networks)
- interpretable and steerable machine learning, or explainable AI (XAI)
- application-driven development of visual analytics systems.
I hold PhD and Bachelor's in Computer Science from Hong Kong University of Science and Technology and Zhejiang University, respectively.
(Full list: Google Scholar)
Interpreting Deep Neural Networks Through Prototype Factorization (Coming soon)
ICDM 2020 DLC; IEEE VIS 2020 Short Paper
Subhajit Das (Intern), Panpan Xu, Zeng Dai, Liu Ren; Junhan Zhao (Intern), Zeng Dai, Panpan Xu, Liu Ren
(TL;DR: model-agnostic method to explain any DNNs with weighted prototypes)
ProtoSteer: Steering Deep Sequence Model with Prototypes
IEEE Transactions on Visualization and Computer Graphics (VAST 2019)
Yao Ming (Intern), Panpan Xu, Furui Cheng, Huamin Qu, Liu Ren
paper (with supplementary materials) video (Dropbox link) bib
(TL;DR: interactive interface to train DNNs with domain knowledge input)
Interpretable and Steerable Sequence Learning via Prototypes
ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD 19), Research Track, Oral
Yao Ming (intern), Panpan Xu, Huamin Qu, Liu Ren
paper (with supplementary materials) video (30MB download) bib
(TL;DR: DNN with built-in interpretability for time-series and event sequence classification)
VisPubData.org: A Metadata Collection about IEEE Visualization (VIS) Publications
IEEE Transactions on Visualization and Computer Graphics 2017
Petra Isenberg, Florian Heimerl, Steffen Koch, Tobias Isenberg, Panpan Xu, Chad Stolper, Michael Sedlmair, Jian Chen, Torsten Möller, John Stasko
paper data visualization demo site
PhD, Computer Science
B.S., Computer Science
"A problem clearly stated is a problem half solved." - Dorothea Brande
"The fundamental principles of designing for people: (1) provide a good conceptual model and (2) make things visible." - The Design of Everyday Things, Don Norman
"The writers who have most to give us often do most violence to our prejudices, particularly if they are our own contemporaries." - Virginia Woolf
Panpan Xu, 2020