Some great news for Taiwo as the spring semester comes to a close.

Taiwo Adebiyi was recognized this week at the Spring Faculty and Staff Meeting of the Cullen College of Engineering (CCE) as the 2025–2026 recipient of the Andrea Prosperetti Research Computing Student Award. This annual award recognizes one graduate student or postdoctoral researcher who has made outstanding contributions to research in scientific computing, data science, high-performance computing, and their applications. Taiwo received the award for his work on TS-roots, a new algorithmic and software framework for efficient and exact Gaussian Process Thompson Sampling in Bayesian optimization.

He was also selected for the 2026–2027 SCIPE Chishiki AI in Civil Engineering Graduate Fellowship, administered through the University of Texas at Austin. Five fellowships are awarded this year, the others going to students from MIT, Johns Hopkins, UCLA, and the University of Georgia. The fellowship provides a $37,000 stipend, up to $12,000 in tuition allowance, travel support, computing access, and mentorship through the Chishiki AI network.

And this summer, Taiwo will join Los Alamos National Laboratory (LANL) as an Advancing Machine Learning for Scientific Discovery (AML) Fellow, a ten-week internship in Los Alamos, New Mexico. Twelve AML fellows are selected nationally, each leading their own project; Taiwo's will be on Bayesian machine learning for stochastic natural gas dynamics. This fellowship will provide a $17,500 stipend as well as travel support.

These recognitions highlight the growing impact of the UQ Lab's work at the intersection of uncertainty quantification, Bayesian optimization, scientific computing, and intelligent engineering systems. Kudos to Taiwo! 🎉

Taiwo has been selected as one of eight recipients of the 2025–2026 UH-Chevron Energy Graduate Fellowship, a prestigious program supporting outstanding graduate students conducting energy-related research at the University of Houston. Each fellow receives a $12,000 award to advance their research, along with mentorship and engagement opportunities with Chevron experts. 🎉

Chosen through a highly competitive process, Taiwo joins a select cohort of fellows across the university recognized for advancing innovative solutions in sustainable and reliable energy systems.

The fellowship is sponsored by Chevron and UH Energy. Similar Chevron Energy Fellowship programs are also offered at The University of Texas at Austin, Texas A&M University, and Rice University.

Congratulations, Taiwo! 🌟

Our latest contribution to Bayesian optimization was presented by Taiwo Adebiyi at the Thirteenth International Conference on Learning Representations (ICLR 2025) in Singapore. As one of the premier conferences in deep learning, ICLR provided a globally recognized platform for the UQ Lab to showcase its research.

Our paper, Optimizing Posterior Samples for Bayesian Optimization by Rootfinding, authored by Taiwo Adebiyi, Dr. Bach Do, and Dr. Ruda Zhang, introduces a highly efficient global optimization algorithm—TS-roots—for Gaussian Process Thompson Sampling (TS) in Bayesian optimization. This is the first method to achieve exact TS, unlocking the full potential of TS’s theoretical advantages for significantly improved Bayesian optimization performance.

TS-roots stands out for its linear scalability with input dimensionality, substantial improvements in inner-loop optimization and competitive outer-loop performance, outperforming several well-known methods including GP-UCB and LogEI. With an average reviewer score of 7.0 placing it in the top 8% of all ICLR 2025 submissions, our paper received considerable attention at the conference.

Building on our strong presence at the NeurIPS 2024 BDU Workshop in Vancouver, ICLR 2025 provided an excellent platform to present the full development of TS-roots, including key algorithmic refinements and new experimental results. To learn more, you can check out the paper, presentation, and GitHub package.

I had the honor of being invited to attend a week-long workshop in Banff, Canada, in February 2025. The workshop on Uncertainty Quantification in Neural Network Models brought together researchers across distinct disciplines to discuss the frontiers of uncertainty quantification (UQ) in modern artificial intelligence models.

The workshop was hosted by the Banff International Research Station (BIRS) for Mathematical Innovation and Discovery, located in the breathtaking (and very cold) Banff National Park. Throughout the week we had engaging discussions—from lectures to coffee breaks, and from the dining hall to the local bar.

This was a truly inspiring research experience, one that will undoubtedly influence the future of UQ research in deep neural nets.

Our UQ@UH Lab proudly made its debut at NeurIPS 2024—the world’s premier conference for artificial intelligence—in Vancouver, Canada.

Our research on Gaussian Process Thompson Sampling via Rootfinding, led by Taiwo Adebiyi, Dr. Bach Do, and Dr. Ruda Zhang, was honored with a contributed oral presentation at the Bayesian Decision-making and Uncertainty (BDU) workshop, placing it among the top 5% of accepted papers.

Taiwo Adebiyi, selected as one of only 17 scholars funded by Google and Meta, led the poster presentation, and engaged with leading experts in Bayesian optimization. Dr. Zhang delivered a well-received oral presentation, showcasing the significance of our findings.

The BDU workshop provided an invaluable platform to engage with global thought leaders, gather insightful feedback, and explore future directions for our research. Exciting developments lie ahead as we continue to push the boundaries of uncertainty quantification and Bayesian optimization!

You can find our presented work on 📄 arXiv and on 💻 GitHub.

Stay tuned for more great work from our team!

Akash attended the 16th World Congress on Computational Mechanics (WCCM) in Vancouver, Canada, from July 21 to 26. He competed as one of six finalists for the UQ-TTA Student Paper Competition in Uncertainty Quantification, where his presentation was well received by the judges. Akash also participated in the general poster session at WCCM and was short-listed for the Best Poster Award. Well done, Akash! 🎉

Toro and Taiwo showcased their skills at the Natural Hazards Engineering Research Infrastructure (NHERI) Computational Academy hosted at the University of Texas, Austin during July 23-26, 2024. And Toro's team won the hackathon! Congratulations! 🎉

This year, our UQ@UH Lab came together for our first-ever Thanksgiving dinner, choosing a local Chinese restaurant in Midtown as the setting for this special occasion. The gathering was particularly meaningful as we welcomed three new members to our lab—Bach, Toro, and Akash—adding to the diverse and vibrant team we’re building.

Given our diverse cultural backgrounds, finding a cuisine that everyone would enjoy could have been a challenge. However, we were pleasantly surprised that Chinese food was a favorite among many of us!

I visited the Isaac Newton Institute for Mathematical Sciences at the University of Cambridge for a week-long summer school program on an Introduction to Uncertainty Quantification in Mechanics of Materials. I was honored to receive a fellowship to participate in this program, which allowed me to meet renowned researchers in the fields of UQ and mechanics of materials. The summer weather in Cambridge was lovely.

Bach Do will join the UQ Lab from Saigon, Vietnam in June 2023. Currently, he is finishing his PhD degree in Architectural Engineering from Kyoto University, Japan. His dissertation focuses on structural design, optimization under uncertainty, and inverse problems in solid mechanics. He received an MS degree from Chulalongkorn University, Thailand, and a BS degree with honors (top 1.5%) from Ho Chi Minh City University of Technology, Vietnam.

We are very eager to welcome Bach here in Houston. But before that, he will have a wedding. Very important indeed.