Topical Group on Data Science

Data science is a fast-growing and highly interdisciplinary field that is at the intersection of statistics, computer science, and mathematics. Applications of data science in engineering and both physical and life sciences are countless and increasing.

News & Announcements

GDS Virtual March Meeting Session #12

Dear GDS Members,
Topical Group on Data Science is hosting its 12th webinar on Friday, June 19, 2020. Please register and join us for another exciting session at the intersection of machine learning and physics.


Title: Scientific Machine Learning for Molecules and Materials


  • 1:00 - 1:05 pm ET | Welcome!
  • 1:05 - 1:40 pm ET | Maxwell Hutchinson - Discovering Materials and Molecules via Active LearningAbstract: Data science, machine learning, and artificial intelligence applications in science and engineering have received rapidly increasing hype over the last several years, with Citrine on the front lines of adoption in materials development. In this talk, I will focus on our most successful approach: data-driven design of experiments, i.e. active learning. Active learning addresses data scarcity, which is a common challenge in scientific applications. Using active learning as a backbone, I will discuss two other techniques for dealing with data scarcity: transfer learning for data reuse and graphical modeling for integrating domain knowledge.
  • 1:40 - 2:15 pm ET | Matthias Rupp - Prediction Errors of Scientific Machine Learning ModelsAbstract: Scientific machine-learning calls for in-depth understanding and control of prediction errors. In this seminar, I will use the context of machine-learning applications in physics, chemistry, and materials science to discuss aspects of prediction errors and their uncertainties, in particular:
  • How to model prediction errors (noise model, predictive distributions)
  • How to assess prediction errors (error metrics for predictive uncertainties, multi-variate stratification, leave-group-out cross-validation)
  • How prediction errors are distributed (scoped errors, input space dependence)

We understand this time may not work for all time zones, but please register even if you are not able to attend. We will record the talks (when possible) and send you a link after the session. You can check out our past webinars here:

Stay connected with GDS - #APSDataScience #gdsvirtual
Become a GDS member
Join our Slack Channel

GDS to Host a Series of Virtual Sessions

GDS is setting up a series of weekly, one-hour webinars for GDS-related March Meeting content to be presented online in the coming months, starting Thursday, March 5th, 2020.

The series will feature a short course on deep learning for image processing, a tutorial on active learning and AI, as well as invited and contributed talks all of which were supposed to take place at the March Meeting.

Register your availability and interest in speaking


Read the GDS Winter 2020 Newsletter

Past News & Announcements

Nominees and award and office holders are expected to meet standards of professional conduct and integrity as described in the APS Ethics Guidelines. Violations of these standards may disqualify people from consideration or lead to revocation of honors or removal from office.