Scientific Machine Learning Workshop

Purdues Discovery Park and the College of Science are hosting a Scientific Machine Learning Workshop for Purdue faculty and research staff. The free event will take place from 9 a.m. to 3 p.m. April 6 in Stewart Center, Room 214.

Machine learning is the ability of a computer system to continue improving its performance based on previous results. In the last decade, as computers have become more powerful and capable, the interest in machine learning-based approaches for science and engineering has grown exponentially, says Toms Daz de la Rubia, chief scientist and executive director of Discovery Park.

Computers can now recognize objects in photos, translate speech and even make recommendations for merchandise to buy or movies to watch — all because they are capable of continuous learning, Daz de la Rubia says. We have only begun to imagine how this kind of technology could revolutionize research.

The U.S. Department of Energys Office of Science has identified 10 priority research directions in machine learning, and its likely that new funding opportunities will emerge to address these priorities, says Patrick J. Wolfe, the Frederick L. Hovde Dean of the College of Science and the Miller Family Professor of Statistics.

Purdue is poised to be a leader in developing new machine learning approaches and in applying machine learning to our research endeavors. To be ahead of the curve, we will create interdisciplinary teams to focus on these research topic areas and new funding opportunities, Wolfe says.

The workshop will provide an overview of the following DOE priority research areas:

* Interpretable machine learning.
* Effective features for scientific machine learning.
* Leveraging domain knowledge and constraints in ML formulations.
* Machine learning in high dimensions.
* Machine learning for enhancing data collection and use on DOE facilities.

* Machine learning for inverse problems and inverse problems for machine learning.
* Reproducibility of machine learning.
* Quantifying the discrepancy in quantities of interest derived using machine learning.
* Machine learning-enabled adaptive scientific computing.
* Addressing the complexity of model architectures and DOE applications.

The event will include presentations by statistics, mathematics and computer science faculty and breakout sessions to discuss potential proposal topics.

Seating for the workshop is limited. Register via Eventbrite by March 30.

For more information, contact Cliff Wojtalewicz, assistant director of Discovery Park, at cliffw@purdue.edu or 765-496-3961.