About
Welcome to my homepage! I’m Huy Nguyen, a third-year Ph.D student at the Department of Statistics and Data Sciences, University of Texas at Austin where I am fortunate to be advised by Professor Nhat Ho and Professor Alessandro Rinaldo. Before that, I graduated from Ho Chi Minh City University of Science with a Bachelor’s degree in Mathematics (Summa Cum Laude). In Summer 2024, I was a research intern at Microsoft AI.
Email: huynm@utexas.edu
Research Interests
My current research focuses on theoretical foundations for the Mixture-of-Experts models. In particular, I investigate the effects of various gating functions (namely the softmax gate, the Top-K sparse softmax gate, the dense-to-sparse gate, the sigmoid gate, etc) on the convergence of expert estimation under the Mixture-of-Experts models. Based on insights from these results, I aim to design novel gating functions and characterize expert networks which help improve the efficiency and scalability of the Mixture-of-Experts applications, including Large Language Models, Multi-modal Learning and Parameter-efficient Fine-Tuning. Additionally, I am also interested in Optimal Transport theory.
(*) denotes equal contribution, (**) denotes equal advising.
Selected Preprints
On Expert Estimation in Hierarchical Mixture of Experts: Beyond Softmax Gating Functions
Huy Nguyen*, Xing Han*, Carl William Harris, Suchia Saria**, Nhat Ho* *
Under review
Revisiting Prefix-tuning: Statistical Benefits of Reparameterization among Prompts
Minh Le*, Chau Nguyen*, Huy Nguyen*, Quyen Tran, Trung Le, Nhat Ho
Under review
Statistical Advantages of Perturbing Cosine Router in Sparse Mixture of Experts
Huy Nguyen, Pedram Akbarian*, Trang Pham*, Trang Nguyen*, Shujian Zhang, Nhat Ho
Under review
Quadratic Gating Functions in Mixture of Experts: A Statistical Insight
Pedram Akbarian*, Huy Nguyen*, Xing Han*, Nhat Ho
Under review
Understanding Expert Structures on Minimax Parameter Estimation in Contaminated Mixture of Experts
Pedram Akbarian*, Huy Nguyen*, Dung Le*, Pedram Akbarian, Nhat Ho
Under review
CompeteSMoE - Effective Training of Sparse Mixture of Experts via Competition
Quang Pham, Giang Do, Huy Nguyen, TrungTin Nguyen, Chenghao Liu, Mina Sartipi, Binh T. Nguyen, Savitha Ramasamy, Xiaoli Li, Steven Hoi, Nhat Ho
Under review
Selected Publications on the Foundations of Mixture of Experts
Sigmoid Gating is More Sample Efficient than Softmax Gating in Mixture of Experts
Huy Nguyen, Nhat Ho**, Alessandro Rinaldo* *
Advances in NeurIPS, 2024.
On Least Square Estimation in Softmax Gating Mixture of Experts
Huy Nguyen, Nhat Ho**, Alessandro Rinaldo* *
Proceedings of the ICML, 2024.
Is Temperature Sample Efficient for Softmax Gaussian Mixture of Experts?
Huy Nguyen, Pedram Akbarian, Nhat Ho
Proceedings of the ICML, 2024.
Demystifying Softmax Gating Function in Gaussian Mixture of Experts
Huy Nguyen, TrungTin Nguyen, Nhat Ho
Advances in NeurIPS, 2023 (Spotlight) .
Statistical Perspective of Top-K Sparse Softmax Gating Mixture of Experts
Huy Nguyen, Pedram Akbarian, Fanqi Yan, Nhat Ho
Proceedings of the ICLR, 2024.
A General Theory for Softmax Gating Multinomial Logistic Mixture of Experts
Huy Nguyen, Pedram Akbarian, TrungTin Nguyen, Nhat Ho
Proceedings of the ICML, 2024.
Towards Convergence Rates for Parameter Estimation in Gaussian-gated Mixture of Experts
Huy Nguyen*, TrungTin Nguyen*, Khai Nguyen, Nhat Ho
In AISTATS, 2024.
On Parameter Estimation in Deviated Gaussian Mixture of Experts
Huy Nguyen, Khai Nguyen, Nhat Ho
In AISTATS, 2024.
Selected Publications on the Applications of Mixture of Experts
FuseMoE: Mixture-of-Experts Transformers for Fleximodal Fusion
Xing Han, Huy Nguyen*, Carl Harris*, Nhat Ho, Suchi Saria
Advances in NeurIPS, 2024.
Mixture of Experts Meets Prompt-Based Continual Learning
Minh Le, An Nguyen*, Huy Nguyen*, Trang Nguyen*, Trang Pham*, Linh Van Ngo, Nhat Ho
Advances in NeurIPS, 2024.
Selected Publications on Optimal Transport
Entropic Gromov-Wasserstein between Gaussian Distributions
Huy Nguyen*, Khang Le*, Dung Le*, Dat Do, Tung Pham, Nhat Ho
Proceedings of the ICML, 2022.
On Multimarginal Partial Optimal Transport: Equivalent Forms and Computational Complexity
Huy Nguyen*, Khang Le*, Khai Nguyen, Tung Pham, Nhat Ho
In AISTATS, 2022.
On Robust Optimal Transport: Computational Complexity and Barycenter Computation
Huy Nguyen*, Khang Le*, Quang Minh Nguyen, Tung Pham, Hung Bui, Nhat Ho
Advances in NeurIPS, 2021.
Recent News
- [Oct 2024] Four new papers on Mixture of Experts are out, [1], [2], [3] and [4].
- [Sep 2024] Three papers on Mixture of Experts, [1], [2] and [3], are accepted to NeurIPS 2024.
- [May 2024] I start my research internship at Microsoft AI where I will work on the applications of Mixture of Experts in Large Language Models.
- [May 2024] Three new papers on Mixture of Experts [1], [2] and [3] are out!
- [May 2024] Three papers on Mixture of Experts, [1], [2] and [3], are accepted to ICML 2024.
- [Apr 2024] I was offered the AISTATS 2024 registration grant. See you in Valencia, Spain this May!
- [Mar 2024] I received the ICLR 2024 Travel Award. See you in Vienna, Austria this May!
- [Feb 2024] Two new papers on the applications of Mixture of Experts in Medical Images [1] and Large Language Models [2] are out!
- [Feb 2024] Two new papers on the theory of Mixture of Experts, [1] and [2], are out!
- [Jan 2024] Two papers on Mixture of Experts, [1] and [2], are accepted to AISTATS 2024.
- [Jan 2024] Our paper “Statistical Perspective of Top-K Sparse Softmax Gating Mixture of Experts” is accepted to ICLR 2024.
- [Dec 2023] Our paper “Fast Approximation of the Generalized Sliced-Wasserstein Distance” is accepted to ICASSP 2024.
- [Oct 2023] I received the NeurIPS 2023 Scholar Award. See you in New Orleans this December!
- [Oct 2023] Our new paper “A General Theory for Softmax Gating Multinomial Logistic Mixture of Experts” is out.
- [Sep 2023] Our new paper “Statistical Perspective of Top-K Sparse Softmax Gating Mixture of Experts” is out.
- [Sep 2023] We have two papers accepted to NeurIPS 2023, [1] as spotlight and [2] as poster.
- [Jul 2023] We will present the paper “Fast Approximation of the Generalized Sliced-Wasserstein Distance” at the Frontier4LCD workshop, ICML 2023.
- [May 2023] Three new papers on the Mixture of Experts theory are out! See more at [1], [2] and [3].
- [Feb 2023] Our new paper on Mixture Models theory “Minimax Optimal Rate for Parameter Estimation in Multivariate Deviated Models” is out.
Professional Services
- Conference Reviewer: ICML (2022,2024), NeurIPS (2022-2024), AISTATS (2022-2025), ICLR (2024-2025) and AAAI (2025).
- Journal Reviewer: Electronic Journal of Statistics.
- Workshop Reviewer: Frontier4LCD (ICML 2023).