I had a wonderful time with my advisor, mentors, and friends at the FoCM conference Vienna, where I presented a poster based on our recent work Function graph transformers universally approximate operators between function spaces. The poster also includes a new application to 2D scalar Helmholtz inversion from sparsely sampled Dirichlet-to-Neumann maps built from partial boundary observations.
I organized a minisymposium with Nick Boffi titled Measure flows for Inverse Problems and Machine Learning (MS35, MS52, and MS68) at the SIAM Conference on Uncertainty Quantification in Minneapolis.
I am thankful to SIAM for supporting my travel.
I just returned from Inverse Days in Helsinki, Finland. This was my second time attending, and both times it has been the highlight of my year.
With the help of my European friends and collaborators, particularly my friend and mentor Matti Lassas, I was able to map out my research program and goals for 2026. These projects will explore new connections between disparate areas of applied mathematics, and I am excited to share them in the coming months.
Just like last year, it was wonderful to reconnect with people I had only met briefly before, and to finally meet in person several people I had known only by name and through their work.
Somehow, I forgot to take even a single picture in Helsinki, so my memories will have to suffice.
I had the pleasure of attending the CBMS Conference on Research at the Interface of Applied Mathematics and Machine Learning (AMML) 2025 at the University of Houston. The highlight of the conference was a ten lecture series by Lars Ruthotto on machine learning and AI. Lars is an outstanding presenter, and these lectures set a new standard for an introduction to machine learning through the lens of applied mathematics. The lecture material and recordings are publicly available.
Additionally, I presented a poster on our ongoing work regarding measure flows in Bayes Hilbert spaces. For this poster, I had some time to experiment with an unorthodox dark mode color scheme. It received positive feedback. The printer did not reproduce the colors exactly as I intended, but I am happy with the result.
The poster was made with Beamer using the Gemini template; you can find my color scheme here. If you are thinking of using it for your own posters, please keep in mind the warnings at the bottom of the README.
This work has evolved into Measure flow path recovery in Bayes Hilbert spaces.
I defended my PhD thesis proposal. Many thanks to my committee Drs. Beatrice Rivière, Anthony Várilly-Alvarado, and Maarten V. de Hoop.
I have just returned from the Applied Inverse Problems conference in beautiful Rio de Janeiro. Many thanks to Matti Lassas, Giovanni Covi, and Siiri Rautio for inviting me to speak in their minisymposium XAIP: Explainable algorithms for inverse problems.
The twelve speakers approached this challenging area from many directions. It is clear that creating interpretable and explainable approaches for scientific machine learning will continue to be one of the most important and difficult challenges in applied mathematics.
I have returned from ICLR in Singapore, where Matti Lassas and I presented our poster on Semialgebraic Neural Networks.
I loved Singapore, and I ate kaya toast several times a day, probably too often at the expense of other great food. But the highlight of the trip was spending time with Matti and Takashi Furuya, who presented multiple papers on measure-valued transformers and neural operators.
I had a fantastic time at Inverse Days in Oulu, Finland, where I presented our work on Semialgebraic Neural Networks. This was my first opportunity to travel internationally for an academic conference, and it was an absolute joy.
It was great to finally meet mathematicians whose work I have read and learned from, including Mikko Salo and Samuli Siltanen. Also, the young researchers I met made me feel welcome beyond anything I expected. I hope this was just the first of many visits to see my friends and collaborators in Finland.
I successfully defended my master's thesis. Many thanks to my committee Drs. Jesse Chan, Anthony Várilly-Alvarado, and Maarten V. de Hoop.