Tri Nguyen
I'm a last-year Ph.D. student working with Prof. Xiao Fu at Oregon State University. I'm interested in understanding and improving machine learning/AI techniques, using tools such as optimization, estimation, identifiability of matrix/tensor decomposition. My soon-to-be dissertation is on noisy label learning. This problem is prevalent across various domains, including clustering, classification, and more recently the LLM alignment using human feedback. My experience working on these different problems through the years suggests that model identification plays a very helpful guidance to design robust solutions with the help of crowd-sourcing labeled data.At the moment, I'm looking for a full-time position in the ML/AI domain, with a focus on the practical applications of machine learning using principles from optimization, estimation, and model identification.
Publications
- [ICASSP2025] T. Nguyen, S. Ibrahim, R. Hutchinson, and X. Fu. Under-Counted Matrix Completion Without Detection Features.
- [NeurIPS2024(spotlight)] T. Nguyen, S. Ibrahim, and X. Fu. Noisy Label Learning with Instance-Dependent Outliers: Identifiability via Crowd Wisdom. [pdf] [code]
- [ICML2023] T. Nguyen, S. Ibrahim, and X. Fu. Deep Clustering with Incomplete Noisy Pairwise Annotations: A Geometric Regularization Approach. [pdf] [code]
- [ICLR2023] S. Ibrahim, T. Nguyen, and X. Fu. Deep Learning From Crowdsourced Labels: Coupled Cross-Entropy Minimization, Identifiability, and Regularization. [pdf] [code]
- [TSP2022] T. Nguyen, X. Fu, and R. Wu. Memory-Efficient Convex Optimization for Self-Dictionary Separable Nonnegative Matrix Factorization: A Frank-Wolfe Approach. [pdf] [code]