Saturday, February 29 and Sunday, March 1
Colorado Convention Center, Room 201

Organizers

Debra J. Audus, National Institute of Standards and Technology
Jonathan K. Whitmer, University of Notre Dame

Sponsors

We thank our generous sponsor: Procter & Gamble

Course Details

Overview

Recent developments in machine learning and related data-driven approaches have created a new paradigm for approaching scientific research. The field of polymer physics has seen important applications in the design of experiments, analysis of scattering data, prediction of molecular properties, and identification of important structural and dynamic patterns. Additionally, the use of high throughput computational and experimental techniques promises to increase the amount of data available to polymer physicists and presents new opportunities for discovery. This day and a half short course will provide an essential introduction to machine learning and data analytics as relevant to polymer physicists, while also showcasing recent advances by leaders in the field. Topics covered will include data capture, design of experiments, varying levels of data quality, model building, optimization and general analysis of both experimental and computational data. Attendees will leave with a sound basis in key algorithmic concepts including when those algorithms are appropriate, an understanding of the state-of-the-art applications, and a foundational understanding of how to incorporate machine learning and data science into their current research.

Who should attend?

The workshop is appropriate for polymer and soft materials researchers at all levels who wish to integrate machine learning techniques into their work. The short course will be particularly useful for people who have not received formal data science training, but appreciate the power of data science to augment and extend traditional techniques. While aimed toward early-career researchers (including graduate students, postdocs, and early career PIs) there will be topics of interest for researchers at all career levels from both computational and experimental groups.

Additional Information

Please bring a laptop for the hands-on tutorials on Saturday. Tutorials will use Google Colab, so no software installation is required beyond access to a web browser. Free wifi will be provided.

Schedule

Saturday, February 29 – Machine Learning Tutorials with Examples

1:00pm - Introduction to Machine Learning – Valentin Stanev (U. Maryland)

2:30pm - Neural Networks I – William Ratcliff (NIST)

3:00pm - Break

3:20pm - Neural Networks II – William Ratcliff (NIST)

4:20pm - Gaussian Processes – Daniel Samarov (NIST)

5:30pm - Natural Language Processing – Debra Audus (NIST)

6:00pm - Wrap up

Sunday, March 1 – Cutting Edge Research

7:45am - Breakfast

8:15am - Machine-learning in nanoscience experiments: X-ray scattering/imaging, autonomous control, and more – Kevin Yager (BNL)

9:15am - Closed-loop experimental design: Theory and Applications – Kristofer Reyes (U. Buffalo)

10:15am - Break

10:35am - Integrating machine learning, simulation, and optimization algorithms for polymer design – David Simmons (U. South Florida)

11:35am - Model Building, Coarse Graining and Free Energy Calculations – Juan de Pablo (U. Chicago)

12:35pm - Lunch

1:30pm - Building predictive models from mixed provenance and accuracy data using transfer learning – Brett Savoie (Purdue Univ.)

2:30pm - Capturing and using experimental data – Cate Brinson (Duke Univ.)

3:30pm - Break

3:50pm - Group Projects

5:15pm - Wrap up