Rhett Olson
he/him
Computer Scientist
he/him
Computer Scientist
I am a computer scientist, currently in the first year of my PhD at the National University of Singapore School of Computing. I am co-advised by Prof. Ng See Kiong and Prof. Roger Zimmermann
In December 2024 I graduated with a Bachelor's of Science in Computer Science and Mathematics from the University of Minnesota College of Science and Engineering
.For six semesters (January 2022-December 2024), I worked as an Undergraduate Teaching Assistant in the UMN Department of Computer Science and Engineering.
As climate change and habitat loss threaten wildlife worldwide, monitoring animal species in natural habitats is a critical problem in ecology and conservation. Passive Acoustic Monitoring (PAM) offers a minimally invasive, low-cost method to monitor species populations by recording audio in natural environments and identifying animal vocalizations in recorded audio. To make PAM efficient and scalable, many machine-learning methods have been proposed for bioacoustic classification to identify animal species from audio of their vocalizations. Our work aims to create global-scale bioacoustic classification models that automatically align predictions with the spatiotemporal context of where and when the audio was recorded. This can reduce the need for transfer learning, which requires expensive manual data annotation to adapt bioacoustic foundation models to the context of local audio datasets.
One work on this topic has been accepted as an applications track paper at ACM SIGSPATIAL 2026. In this work, we present Spatiotemporal Mixture of Regional Experts (SMoRE), the first global-scale bioacoustic species classifier jointly trained on both audio and spatiotemporal context. This work significantly improves performance of classifiers directly applied to new bioacoustic data, without any prior knowledge or annotated labels for each evaluation setting, reducing the need for data annotation and transfer learning in PAM systems.
Historical maps contain rich information about the past, and serve as vital sources for scholars in many different fields. Through this project, I have developed a novel approach to facilitate scholarly analysis of place names on historical maps. This approach enables scholars to conduct queries for instances of a given place (e.g. "Istanbul") that retrieve dated, scanned historical maps of that place, which reveal when different names for that place were used over time (e.g. when "Istanbul", "Constantinople", and "Byzantium" were used on many different maps). Read more
As a participant in the 2024 DIMACS REU Program hosted by Rutgers University, I research the problem of truth learning in social networks. Specifically we aim to characterize when it is possible for large networks to collectiely learn some truth with high probability, despite the presence of adversarial agent who try to spread misinformation. Read more
For my work on the above two projects, I was named an Honorable Mention for the 2024-2025 CRA Outstanding Undergraduate Researcher Awards Program.
For this paper, I won 3rd prize in the undergraduate category of the 2024 ACM Student Research Competition Grand Finals, and 1st prize in the ACM SIGSPATIAL Student Research Competition.
Check out my CV! (updated September 21st, 2026)
I made an adversarial version of Pile-Up-Poker, which extends the poker-based puzzle game from Puzzmo to challenge the player to construct strong poker hands in a 6x6 grids, while the CPU adversarially places one card in the grid each round to try to minimize the player's score. Please give it a try, and feel free to get in touch if you have any feedback!