September 9, 2019, London, UK

Combining Physical and Data-Driven Knowledge in Ubiquitous Computing

Ubicomp 2019 Workshop


Real-world ubiquitous computing systems face the challenge of requiring a significant amount of data to obtain accurate information through pure data-driven approaches. The performance of these data-driven systems greatly depends on the quantity and `quality' of data. In ideal conditions, pure data-driven methods perform well due to the abundance of data. However, in real-world systems, collecting data can be costly or impossible due to practical limitations. Physical knowledge, on the other hand, can be used to alleviate these issues of data limitation. This physical knowledge can include domain knowledge from experts, heuristics from experiences, as well as analytic models of the physical phenomena.

This workshop aims to explore the intersection between (and the combination of) data and physical knowledge. The workshop will bring together domain experts that explore the physical understanding of the data, practitioners that develop systems and the researchers in traditional data-driven domains. The workshop welcomes papers addressing these issues in different applications/domains as well as algorithmic and systematic approaches to apply physical knowledge. Therefore, we further seek to develop a community that systematically analyzes the data quality regarding inference and evaluates the improvements from the physical knowledge. Preliminary and on-going work are welcomed.

Please check our CPD-18 workshop here.


Many real-world ubiquitous computing systems make use of data-driven algorithms that require a significant amount of data to obtain good performance. System performance of these pure, data-driven systems largely depends on the quantity and quality of the data they use. Under ideal conditions – representative, large, balanced and labeled data – pure data-driven methods perform very well. However, in real-world systems, collecting data can be costly or impossible due to practical limitations. Physical knowledge (strong priors), can help alleviate issues that emerge when good data is limited. This includes 1) domain knowledge from experts, 2) experience-driven heuristics, and 3) analytic models of physical phenomena. With physical knowledge, we can infer target information more accurately compared to purely data-driven models. These priors can also improve performance and robustness when limited labeled data is available. In recent years, researchers have combined physical knowledge with traditional, data-driven approaches to improve model accuracy and system performance. We aim to attract researchers that are exploring fundamental questions about the integration of physical knowledge and data in real-world systems and deployments. We also aim to identify solutions and methodologies that generalize across various application domains.

Topics of Interests

  • - Innovations in learning algorithms that combine physical knowledge or models for sensor perception and understanding
  • - Experiences, challenges, analysis, and comparisons of sensor data in terms of its physical properties
  • - Sensor data processing to improve learning accuracy
  • - Machine learning and deep learning with physical knowledge on sensor data
  • - Mobile and pervasive systems that utilize physical knowledge to enhance data acquisition
  • - System services such as time and location estimation enhanced by additional physical knowledge
  • - Heterogeneous collaborative sensing based on physical rules

The application areas include but not limited to:

  • - Human-centric sensing applications
  • - Environmental and structural monitoring
  • - Smart cities and urban health
  • - Health, wellness and medical

Successful submissions will explain why the topic is relevant to the data limitation caused problem that may be solved through the physical understanding of domain knowledge. In addition to citing relevant, published work, authors must cite and relate their submissions to relevant prior publications of their own. Ethical approval for experiments with human subjects should be demonstrated as part of the submission.

Important Dates

Submission Deadline: June 23, 2019 June 30, 2019, submit HERE

Author Notifications: July 1, 2019 July 8, 2019

Camera-ready: July 8, 2019 July 12, 2019

Workshop Date: September 9 and 10, 2019

Submission Guidelines

Please submit short papers using the SIGCHI Extended Abstract format with 2-8 pages of content. Note that we use the newest ACM SIGCHI Extended Abstract template. Please refer to this link. Submissions may include as many pages as needed for references. The submissions should not be anonymous.

Template download: here


Workshop Chairs

Xinlei Chen Carnegie Mellon University

Shijia Pan Carnegie Mellon University

Jorge Ortiz Rutgers University

Advising Committee

Rasit Eskicioglu University of Manitoba

Pei Zhang Carnegie Mellon University

Hae Young Noh Carnegie Mellon University

Jie Liu IEEE Fellow Microsoft Research Harbin Institute Technology

Pan Hui IEEE Fellow University of Helsinki Hongkong University of Science and Technology

Technical Programm Committee

Yong Li Tsinghua University

Roozbeh Jafari Texas A&M University

Mi Zhang Michigan State University

Yuan Tian University of Virginia

Yanjun Han Stanford University

Bing Liu Facebook AI

Ming Zeng Facebook Inc.

Pan Hu Stanford University

Yong Zhuang Carnegie Mellon University

Xiaoxuan Lu Oxford University

Jonathon Fagert Carnegie Mellon University

Mostafa Mirshekari Carnegie Mellon University

Jun Han National University of Singapore

Dezhi Hong University of California San Diego

Wen Hu The University of New South Wales


PoI as the Physical Knowledge for Human Behavior Prediction

Yong Li. Associate Professor of the Department of Electronic Engineering, Tsinghua University


The CPD 2019 workshop is part of (co-located with) Ubicomp 2019, which will be held at London.