Angela Zhu

Angela Zhu

PhD student at OSU

Academic Works

Journal Articles

Exploring MCMC Samplers for N-Mixture Models

Published in 2026 Confluence Symposium at Oregon State University, 2026

The goal of this work is to learn the mechanistic behavior of Markov Chain Monte Carlo (MCMC) samplings algorithms used for N-Mixture models. We develop our own MCMC samplers using the Metropolis Hastings sampling algorithm and compare against NIMBLE. We start by creating simulated data and then use our samplers to learn back the parameter estimates. To be clear, this is not novel research, this was a project to understand how mcmc sampling works for an ecologial scenario.

Project was the final project of a graduate Advanced Probabilistic Graphical Models course (AI 539) at Oregon State University taught by Professor Weng-Keen Wong, artifacts include a presentation and a paper to document the work. And a poster was presented as apart of the Confluence Symposium held at OSU.

paper poster presentation code

Investigating Different Geo Priors for Image Classification

Published in Fine-Grained Visual Categorization (FGVC) workshop at CVPR, 2025

I investigated the effectiveness of SINR as a geographical prior for an image classification task and the performance implications of combining the outputs of multiple predictive models. This work was accepted to FCVC 12 workshop at CVPR.

Independent study for the Spring 2025 term supervised by Professor Subhransu Maji and in collaboration with the rest of the GeoModel Research Group.

paper poster

DS4CG iNaturalist Geomodel Annotation Tool

Published in Center for Data Science and New England Computer Vision (NECV) Workshop, 2024

Expert-in-the-loop platform to collect species data and improve species distribution model accuracy. Developed and deployed a web application that enables experts to annotate species range maps using predictions from the SINR GeoModel- ML-based range prediction model to refine species distributions.

Project was apart of the Data Science for the Common Good Program at UMass Amherst in collaboration with UMass Researchers Prof. Grant Van Horn and Prof. Subhransu Maji.

post video code poster