Kevin Shuai Xu
Kevin S. Xu

Assistant Professor, EECS Department
University of Toledo
2801 W. Bancroft St. MS 308
Toledo, OH  43606-3390, USA



Recent news

2019

  • Dec. 2: I will be teaching two special topics courses in Spring 2020: EECS 4980/5980 (Social and Information Networks) and EECS 6980/8980 (Probabilistic Methods in Data Science). Tentative list of topics
  • Sept. 16: Our recent grant proposal on predicting kidney transplant survival has been funded through the Joint DMS/NLM Initiative on Generalizable Data Science Methods for Biomedical Research. Thanks to the NIH/NLM and NSF for supporting our research! Link to award abstract
  • Apr. 1: Fixed dead links for AFFECT and regularized graph layout MATLAB toolboxes.
  • Feb. 25: Our paper "The Block Point Process Model for Continuous-Time Event-Based Dynamic Networks" has been accepted to WWW 2019. A preprint is available at https://arxiv.org/abs/1711.10967

2018

  • Oct. 27: I recently gave a talk in the University of Michigan MIDAS Seminar series on statistical models for analyzing dynamic social network data: Slides Video
  • Aug. 21: Prerequisites for my course EECS 5750 Machine Learning are incorrectly listed. Graduate students interested in enrolling for Fall 2018 but unable to due to the prerequisite error can email me to obtain an override.
  • Jul. 13: I recently gave a tutorial "Generative Models for Social Media Analytics: Networks, Text, and Time" at ICWSM 2018 along with Jimmy Foulds. Tutorial materials (presentation slides and Python code for demos)
  • Apr. 23: Our paper "Leveraging Friendship Networks for Dynamic Link Prediction in Social Interaction Networks" has been accepted to ICWSM 2018. A preprint is available at https://arxiv.org/abs/1804.08584
  • Mar. 27: I have received the NSF CISE Research Initiation Initiative (CRII) award. Thank you to the NSF for supporting my research! Link to award abstract
  • Jan. 16: Updated dynamic stochastic block models MATLAB toolbox with implementations of the stochastic block transition model. The toolbox is now hosted on Github: Link to toolbox
  • Jan. 12: I will be teaching two special topics courses this semester: EECS 4980/5980 (Social and Information Networks) and EECS 6980/8980 (Probabilistic Methods in Data Science). Tentative list of topics

2017

  • Oct. 3: Our recent research using physiological data collected from wearables to predict autonomic nervous system states is featured in the IEEE Xplore Innovation Spotlight.
  • Sept. 17: Presented our paper "Unsupervised motion artifact detection in wrist-measured electrodermal activity data" (co-authored with Yuning Zhang and Maysam Haghdan) at the International Symposium on Wearable Computers (ISWC) 2017: Preprint Slides
  • Aug. 18: One of the listed prerequisites for my course EECS 4750/5750 Machine Learning is out of date. Students interested in enrolling for Fall 2017 but unable to due to the prerequisite error can email me to obtain an override.
  • Jan. 31: Our paper "A Compressed Sensing Based Decomposition of Electrodermal Activity Signals" has been accepted to the IEEE Transactions on Biomedical Engineering. A preprint is available at https://arxiv.org/abs/1602.07754

Research

I direct the Interdisciplinary Data Engineering and Science (IDEAS) Laboratory at the University of Toledo. Our research is primarily in the areas of machine learning and statistical signal processing with applications to network science, human dynamics, and health.

Current projects

  • Creation of statistical models and efficient inference procedures for discrete-time and continuous-time dynamic network data, particularly social network data (supported by the National Science Foundation and National Geospatial Intelligence Agency)
  • Prediction of survival times for kidney transplants by integrating high-dimensional survival analysis and network models with biological models of immunogenicity (supported by the National Institutes of Health and the National Science Foundation)
  • Development of robust algorithms for analysis of physiological data (including electrodermal activity and heart rate) collected using wearable sensors (supported by the University of Toledo Research Awards and Fellowship program)
  • Prediction of chronic PTSD from fMRI brain imaging data (supported by the National Institutes of Health)

Past projects

  • Prediction of human performance and cognitive load levels for human-machine collaboration (supported by the Ohio Federal Research Network)

Education

  • PhD, Electrical Engineering: Systems, University of Michigan, 2012, Advisor: Alfred O. Hero III
  • MSE, Electrical Engineering: Systems, University of Michigan, 2009
  • BASc, Honors Electrical Engineering, University of Waterloo, 2007

Honors and awards

  • NSF Computer and Information Science and Engineering (CISE) Research Initiation Initiative (CRII) Award, 2018
  • Exceptional Service Award, International Conference on Social Computing, Behavioral-Cultural Modeling, & Prediction and Behavior Representation in Modeling and Simulation (SBP-BRiMS), 2016
  • Winner, International Conference on Social Computing, Behavioral-Cultural Modeling, and Prediction Challenge (SBP), 2013
  • Postgraduate scholarship: Doctorate, Natural Sciences and Engineering Research Council of Canada (NSERC), 2010-2011
  • Postgraduate scholarship: Master’s, Natural Sciences and Engineering Research Council of Canada (NSERC), 2008

Selected publications

For a full publication list, please see my CV or my Google Scholar Citations profile.
  • Junuthula, R. R., Haghdan, M., Xu, K. S., & Devabhaktuni, V. K. (2019). The Block Point Process Model for continuous-time event-based dynamic networks. In Proceedings of the World Wide Web Conference (pp. 829-839). arXiv: 1711.10967
  • Zhang, Y., Haghdan, M., & Xu, K. S. (2017). Unsupervised motion artifact detection in wrist-measured electrodermal activity data. In Proceedings of the 21st ACM International Symposium on Wearable Computers (pp. 54–57). arXiv:1707.08287
  • Jain, S., Oswal, U., Xu, K. S., Eriksson, B., & Haupt, J. (2017). A compressed sensing based decomposition of electrodermal activity signals. IEEE Transactions on Biomedical Engineering, 64(9), 2142–2151. arXiv:1602.07754
  • Han, Q., Xu, K. S., & Airoldi, E. M. (2015). Consistent estimation of dynamic and multi-layer block models. In Proceedings of the 32nd International Conference on Machine Learning (pp. 1511–1520). arXiv:1410.8597
  • Xu, K. S. (2015). Stochastic block transition models for dynamic networks. In Proceedings of the 18th International Conference on Artificial Intelligence and Statistics (pp. 1079–1087). arXiv:1411.5404
  • Xu, K. S., & Hero III, A. O. (2014). Dynamic stochastic blockmodels for time-evolving social networks. IEEE Journal of Selected Topics in Signal Processing, 8(4), 552–562. arXiv:1403.0921.
  • Xu, K. S., Kliger, M., & Hero III, A. O. (2014). Adaptive evolutionary clustering. Data Mining and Knowledge Discovery, 28(2), 304–336. arXiv:1104.1990.

Teaching

  • EECS 1100 Digital Logic Design: Fall 2019, Fall 2018, Fall 2017
  • EECS 1510 Introduction to Object-Oriented Programming: Fall 2015
  • EECS 4750/5750 Machine Learning: Fall 2019, Fall 2018, Fall 2017, Fall 2016
  • EECS 4980/5980 Special Topics – Social and Information Networks: Spring 2020, Spring 2018
  • EECS 6980/8980 Special Topics – Probabilistic Methods in Data Science: Spring 2020, Spring 2018
  • EECS 6980/8980 Special Topics – Social Network Analysis: Spring 2016

Selected service to research community

Software

Software packages resulting from my research can be found at the IDEAS Lab GitHub page. Older software packages:

Recent presentations

  • Statistical models for analyzing dynamic social network data: MIDAS Seminar (University of Michigan) 2018 Slides Video
  • Unsupervised motion artifact detection in wrist-measured electrodermal activity data: ISWC 2017 Slides
  • The block point process model for continuous-time event-based dynamic networks: JSM 2017 Slides
  • Consistent estimation of dynamic and multi-layer block models: ICML 2015 Slides Video
  • Stochastic block transition models for dynamic networks: AISTATS 2015 Slides

Curriculum vitae