• Two Fellowships at $35,000 each
• Six Fellowships at $20,000 each
• Fifty Fellowships at $8,500 each

Machine Learning for Scientific Computing 

Felix Ye, SUNY Albany

12-1pm EST Monday, Feb. 3, 2025
409 Sondheim Hall, UMBC

 In the first part, I will introduce a nonlinear stochastic model reduction technique for high-dimensional stochastic dynamical systems that have a low-dimensional invariant effective manifold with slow dynamics and high-dimensional, large fast modes. Given only access to a black-box simulator from which short bursts of simulation can be obtained, we design an algorithm that outputs an estimate of the invariant manifold, a process of the effective stochastic dynamics on it, which has averaged out the fast modes, and a simulator thereof. This simulator is efficient in that it exploits of the low dimension of the invariant manifold, and takes time-steps of size dependent on the regularity of the effective process, and therefore typically much larger than that of the original simulator, which had to resolve the fast modes. The algorithm and the estimation can be performed on the fly, leading to efficient exploration of the effective state space, without losing consistency with the underlying dynamics. 

The second part focuses on optimal transport (OT), a powerful framework for comparing probability distributions. Applications such as shuffled regression can be approached by optimizing regularized optimal transport (OT) distances, such as the entropic OT and Sinkhorn distances. A common approach for this optimization is to use a first-order optimizer, which requires the gradient of the OT distance. For faster convergence, one might also resort to a second-order optimizer, which additionally requires the Hessian. The computations of these derivatives are crucial for efficient and accurate optimization. However, they present significant challenges in terms of memory consumption and numerical instability, especially for large datasets and small regularization strengths. We circumvent these issues by analytically computing the gradients for OT distances and the Hessian for the entropic OT distance, which was not previously used due to intricate tensor-wise calculations and the complex dependency on parameters within the bi-level loss function. Through analytical derivation and spectral analysis, we identify and resolve the numerical instability caused by the singularity and ill-posedness of a key linear system. Consequently, we achieve scalable and stable computation of the Hessian, enabling the implementation of the stochastic gradient descent (SGD)-Newton methods. 

Felix Ye is a Assistant Professor in the Department of Mathematics and Statistics at SUNY Albany. His research interest is the intersection of machine learning and dynamical systems and is directed toward data-driven model reduction methods in the context of stochastic dynamical systems. He was a Postdoctoral Fellow at Johns Hopkins University and received a PhD in applied math from the University of Washington, advised by Hong Qian.

• Guided tours of Oxford and Birmingham, birthplace of the Industrial Revolution
• Visit Sarehole Mill, Tolkien’s childhood village and inspiration for The Shire, and The Kilns, Lewis’ home outside Oxford
• Tickets to Oxford’s Science Museum, Botanical Garden, Museum of Natural History, Pitt Rivers Museum, and Bodleian Library
• Live and study in University of Oxford residence halls and classrooms

Dear campus partners:

Our training series on navigating immigration processes will resume this month!

Overview: The training series will cover topics related to the J-1 and H-1B visa categories, ensuring that you are well-equipped to support our international faculty, staff, and scholars. The sessions will provide practical insights, address common challenges, and foster a deeper understanding of immigration in higher education.

Who: 

- The intended audience is both seasoned and new staff in academic departments or research centers, who are involved in hiring, onboarding, and terminating international employees in paid or unpaid positions. 

- Anyone who would like to learn about and understand our field are also welcome, just be ready for an in-depth look at immigration items.

- Shared Services staff are invited to attend sessions as well.

Key Focus Areas:

• Visa Categories and Processes: An in-depth exploration of the H-1B and J-1 visa categories and the associated processes.

• Immigration Compliance: Understanding the legal aspects of immigration to ensure UMBC's compliance with government regulations.

• Navigating Changes: Learning when to communicate changes to our office.

Location: in-person in Admin 218 and simultaneously via Webex, the link is here.

Registration: please email me at dianezg@umbc.edu and I will add the events to your calendars. Then, please RSVP YES directly in the Google calendar.

Schedule: All sessions are from 3 pm to 4 pm, simultaneously in Admin 218 and virtually, unless noted. Please note that you do not have to attend all the sessions.

We encourage all staff involved in immigration-related processes to participate actively in this series. Your engagement is pivotal in ensuring a seamless and supportive environment for our international faculty, researchers, and staff.

We look forward to your active participation in this valuable training initiative.

The dates for 2025 are:

Tuesday, February 18

Tuesday, March 11

Tuesday, April 15 - Canceled

Tuesday, May 20

Friday, June 13, from 10am-3pm, full day of training with lunch. Topics will include J-1, H-1B, and F-1 students. Register soon by emailing dianezg@umbc.edu!

Tuesday, September 16

Tuesday, October 21

Tuesday, November 18

Tuesday, December 16

Location: Admin 218 or Webex link here.

I hope to see you very soon!

Diane Zeenny Ghorayeb

International Scholar Coordinator, ISSS, Center for Global Engagement

The Barbara E. Burkman Public Health Scholarship Endowment Fund scholarship application is now available in Scholarship Retriever.  

Scholarship Application Deadline: February 14th, 2025

Please follow the link below to Scholarship Retriever where you can review the scholarship criteria and complete your application.

Scholarship Retriever

https://scholarships.umbc.edu/retriever/  

Barbara E. Burkman Scholarship

If you have any questions please contact Melissa Cox - mrose4@umbc.edu