Benjamin A. Burns

Hello! I am a third-year Machine Learning PhD student at Georgia Tech advised by Sara Fridovich-Keil. My research interests are in generative modeling and in inverse problems. I am grateful to be supported by the NSF Graduate Research Fellowship.

Before moving to Georgia Tech, I completed my undergrad in math and CS at UMass Amherst, where I had the privilege to work with Markos Katsoulakis and Benjamin Zhang on score-based diffusion models.

I am passionate about teaching, with an emphasis on high school and early-college STEM education. During my time at UMass, I was heavily involved with the Undergraduate Course Assistant (UCA) program, serving both as a Head UCA and as a UCA program coordinator. I am proud to have been recognized with one of the inaugural Lifetime UCA awards.

publications

B. A. Burns, S. Fridovich-Keil. When, why, and how do diffusion posterior samplers fail? A finite-sample lens. 05/2026.

* denotes equal contribution, (α-β) denotes alphabetical author ordering.

recently

[2026-10-19]I was awarded the Steve W. Chaddick Fellowship. [2026-10-16]I presented two posters at the ML@GT Student Conference on our papers "When, Why, How" and PC-DAPS (forthcoming). [2026-05-28]New preprint on understanding when, why, and how diffusion posterior samplers fail is available on arXiv, joint with Sara Fridovich-Keil. [2026-04-03]I presented our on-going work on SVGD with neural operators at the 2026 Southeast ACM Student Workshop at Emory University. [2026-03-25]I presented our on-going work on SVGD with neural operators in MS160 at SIAM UQ26. [prehistory]old news

some writings

[2024-12-27]Picking UMass CICS Electives [2024-07-15]Navigating the UMass CICS course requirements [2024-05-21]Advice for UMass Math + CS undergrads interested in research [2024-04-21]My UMassCTF 2024 challenges [2023-06-25]"Missing Skills" not taught by the UMass CICS curriculum